Commons:Lingua Libre/Chat room/2026

Welcome to the Chat room!

Place used to discuss any and all aspects of Lingua Libre: the project itself, discussions of the operations, policy and proposals, technical issues, etc. For code-oriented issues visit Commons:Lingua Libre/Technical board.

Feel free to participate in any language you want to.

Start a new discussion

Chatroom FAQ

Ongoing: Following our May 2026 full system upgrade, this FAQ section and most of our wikipages require to be reviewed. Content and links may be obsolete. You may help by updating them. ― User:Yug

What IRL events are coming? When? Where?

Let Wikimedians know about your Lingua Libre events : add yours on Lingua Libre/Events.

How to translate LinguaLibre in your language ?

When on Lingualibre.org, click on the top-bar « Edit » icon . Once on translatewiki.net, select your language and add your micro-translations.

Volunteers activate new translations each month. If late, please message us on the forum.

How to add missing languages?

Natively supported languages have been expanded from 700 to 10,000+ languages, dialects and creoles. If your language is missing, please submit your request on Commons talk:Lingua Libre, you must provide 3 elements : the wikidata Qid, name, and ISO 639-3 code. For now, volunteers will then add your language into gitlab's languages-by-qid.json. (Full tutorial to write in Help:Lingua Libre/Add a language.)

How to keep my wikimedia project up to date?

Contact WikiLucas00, the botmaster of Lingua Libre Bot. For more info, check out Help:Lingua Libre/Bots and Lingua Libre/Bot.

How to download all audios of one language? By speaker?

A script is available at CommonsDownloadTool. For more, see Help:Lingua Libre/Download datasets.

How to archive sections which have been answered?

After reviewing the section, add {{done}} ~~~~ to the top of the section. After few days to 2 weeks, move the section's code to [[Commons:Lingua Libre/Chat_room/Archives/{YYYY}]].

Opening Lingua Libre project on Commons

Welcome ! We are soon opening the Lingua Libre project on Wikimedia Commons. This wiki project is the local support and data for files within the Category:Lingua Libre. We created and therefore care for over 1.5 millions audios files in +300 languages. Yug (talk) 12:36, 9 October 2025 (UTC)reply

Project structure

Commons page	Role	Comment
Commons:Lingua Libre/AppNotice	i18n community notification for users of LinguaLibre.org.	front-end/src/components/RibbonNotice.vue
Commons:Lingua_Libre/SiteNotice	i18n community notification for users of the recording studio (Lingua Libre Django).	/src/components/RibbonNotice.vue
Commons:Lingua_Libre/List/{iso}/{Title}	Words or text lists in quick access in Lingua Libre recorder's List step.
Category:Lingua Libre	Root category for the project's assets on Commons, audio, visuals, presentations, events photos, etc.
Category:Lingua Libre pronunciation	Core category for all our recordings.	To rename "Lingua Libre recordings"
{{Lingua Libre help}} & {{Lingua Libre technical}}	Navigation template for Lingua Libre project on Commons and beyond.

Announcement : New recording Studio pre-release !

Done. Yug (talk) 20:33, 31 May 2026 (UTC)reply

See message here Yug (talk) 11:16, 16 October 2025 (UTC)reply

Hi team. Absolutely loving the new version! I love that I don't have to switch into another language version of the display to add words in that language. I am in the situation where I can speak words from other languages but I am not fluent in the written language. I think this will make it a lot easier for folks to donate their voices to the project.

I did have a few questions about the profile tab. It asks for location multiple times (after each language is added, and at the end of the profile). Is it a mechanism to connect the audio to specific accents from certain locations? For example, I learnt to speak on the south coast of England, but have lived in Scotland for more than 20 years. Or is it just the current physical location of the speaker? If the profile is edited after audio files are uploaded, are the changes applied retrospectively to previously recorded content? Any help would be deeply appreciated.

Drkirstyross (talk) 12:29, 10 December 2025 (UTC)reply

Hello Drkirstyross,

I missed your question about the language's location at first. Your inference is correct. By example I learnt my Chinese in Taiwan, and within Lingua Libre I need to specify where I learnt my Chinese: in Taiwan. It gives an strong hint of your accent therefor.

The place of residence (the last place requested) is either you current or your most influential place of residence when it comes to your language learning.

And at the end yes, i do have a strong Taiwanese accent when i speak Chinese, as well as a Strong French accents, matching my overall place of residence. --Yug (talk) 14:18, 11 June 2026 (UTC)reply

Hi @Yug. Brilliant. Thank you for the reply. If I fix my profile to reflect this, will these changes duplicate to content I previously recorded? Or would I need to pop onto Commons and modify a metadata property? I want to make sure they are as useful as possible to the projects. Drkirstyross (talk) 08:21, 13 June 2026 (UTC)reply

Hello Drkirstyross,

Your language(s) location(s) are attached to your profile, and therefore only indirectly to your audio. You do ́not need to edits your files on Commons. For now this data is rather hidden, but we may make it more available in the future if we assess that General Data Protection Regulation (Regulation (EU) 2016/679) privacy laws allows us to do so. Yug (talk) 21:11, 14 June 2026 (UTC)reply

Hi @Yug. Brilliant. Thanks for letting me know. It is sincerely appreciated. Drkirstyross (talk) 18:30, 15 June 2026 (UTC)reply

fra -> sas correction needed

Done (by myself). Yug (talk) 20:33, 31 May 2026 (UTC)reply

Hello Pitchrigi, Hzu Hzu, Elmawardana20 and Rina Harpani, Your recorded the following files in French. You want them in LL-Q1294047 (sas), right ?

Yug (talk) 09:47, 15 November 2025 (UTC)reply

Reporting bugs

Welcome to this "Reporting bugs" section for early testers of the new Lingua Libre !

A Temporary demonstrator is secretly available to early testers at https://lingua-libre.toolforge.org . We invite testers to use various devices and report bugs here or better, on our Phabricator workboard. Please share your device name, browser name, version if known, and your Operating System, ex: Asus K40J, Firefox 143.0.4 Ubuntu 24.04.

As we are heading to put this Alpha into production we will mostly list bugs, fix the most notable ones, but we won't move elements around for optimization nor add feature for now. Those requests will be assessed for later coding cycles. You may use the following template:

=== Bug(s) report by {username} ===
* '''Bug name:''' {give it a name}
* '''Place:''' Step {n} > element {name, ex: "Test microphone!"}
* '''Incorrect behavior:''' {describe incorrect behavior}
* '''Wanted behavior:''' {describe}
* '''Device:'''
* '''Web browser:'''
* '''Phabricator task:''' Yes/No (if yes, share the task name)

We hope you will enjoy this. Yug (talk) 09:56, 16 October 2025 (UTC)reply

Bug reported by djknusper

Bug name: Progressbar does not show progress during upload.
Place: Step 6 > Progressbar
Incorrect behavior: It fills up quickly full, but many files are still to be uploaded.
Wanted behavior: Unsure, either it should display the upload progress of each individual soundfile or it should show the progress of the overall upload. Right now it is just stuck and full blue, but the upload is working.
Device: Laptop
Web browser: Firefox 147.03

Hello User:Pushkar7077, just letting you know about this Step 6 progress bar issue. I shall create a Phabricator soon on that. Yug (talk) 20:33, 31 May 2026 (UTC)reply

Bug(s) report by Yug

Done. Code fixed. Yug (talk) 20:33, 31 May 2026 (UTC)reply

Bug name: Double SDC on old records.
Place: Step 6 > write on the file's SDC.
Incorrect behavior: 's SDC is doubled.
Wanted behavior: The condition to wipe out the SDC.
Device: n.a.
Web browser: n.a.
Phabricator task: Yes/No (if yes, share the task name)

Done Yug (talk) 14:16, 4 November 2025 (UTC)reply

Hello. That was my recorded audio, I just tested it.

Device is Asus VivoBook X515FA, web browser is Chrome. But I don't understand 'incorrect behavior'? Inosham (talk). 12:15, 18 November 2025 (UTC)reply

Hello Inosham , nothing from your side. The code behave the wrong way so I fixed it. Yug (talk) 20:33, 31 May 2026 (UTC)reply

Bug report by Lucas Werkmeister

Done. Yug (talk) 20:33, 31 May 2026 (UTC)reply

Category:Lingua Libre pronunciation with incomplete parameters is filling up at an alarming rate, and a lot of the files in there (random example) have invalid wikitext (leading whitespace resulting in == Summary == instead of a real heading). Please take a look. --Lucas Werkmeister (talk) 14:13, 1 November 2025 (UTC)reply

I think the “incomplete parameters” category might be there because new files are getting point in time (P585) instead of recording date (P10135)? Lucas Werkmeister (talk) 15:32, 1 November 2025 (UTC)reply

⇒ !92 Lucas Werkmeister (talk) 18:01, 1 November 2025 (UTC)reply

@Yug: Ping? Lucas Werkmeister (talk) 13:19, 4 November 2025 (UTC)reply

@Lucas Werkmeister, thanks for the ping I missed this one ! Indeed, damn

I fixed the == Summary == a few days ago. Let me inspect the other stuff. Yug (talk) 14:15, 4 November 2025 (UTC)reply

Ok! Found it ! This diff removes the Category:Lingua Libre pronunciation with incomplete parameters. So, we have to make the SpeakerId: {Lingualibre.org Qid} optional from {{Lingua Libre record}}. Indeed, this info is phased out with the new App release. Note: I don't have the rights to edit the module which powers this template. Yug (talk) 14:35, 4 November 2025 (UTC)reply

9ef150 Indentation code to fix
!92 P585->P10135 code fix. On toolforge, I dit git pull then restarted the server. File:LL-Q150_(fra)-Yug-lynx.wav#P10135 Date of recording seems to work fine now. Thanks Lucas !
Fix indentation error in recent uploads, thanks to Special:Contributions/Dragons_Bot !
Done Fix Module:Lingua_Libre_record#L-268 for Category:Lingua Libre pronunciation with incomplete parameters: needs to be edited, SpeakerId is not required anymore, remove that condition.

We are back on track. Yug (talk) 18:09, 4 November 2025 (UTC)reply

Suggested fix: Special:Diff/1109106214; I’ll probably deploy it to the non-sandbox module later tonight if nobody objects. Lucas Werkmeister (talk) 18:49, 4 November 2025 (UTC)reply

Done Lucas Werkmeister (talk) 20:40, 4 November 2025 (UTC)reply

Good teamwork ! Yug (talk) 21:04, 4 November 2025 (UTC)reply

Bug(s) report by Yug (2)

Bug name: Empty SDC
Place: ~1,000 files in Category:Lingua Libre pronunciation with incomplete parameters
Incorrect behavior: Some upload by the new app had empty structured data on commons
Wanted behavior: Fill back the SDC
Device: n.a.
Web browser: n.a.
Phabricator task: None for now.

The list of users affected :

Benediktus Ryan (talk · contribs) (285),
Swarabakti (talk · contribs) (15),
Amidaxaviera (talk · contribs) (2),
Antya Anantari (talk · contribs) (32),
Akhmadraflii (talk · contribs) (20)
Nada Auliya (talk · contribs) (19),
Empat Tilda (talk · contribs) (12),
Heni say (talk · contribs) (43),
Sizzling Tea (talk · contribs) (36),
Intannx6424 (talk · contribs) (25),
Maulana Fardan Alhamdi (talk · contribs) (22),
Aibaeys (talk · contribs) (25)

Yug (talk) 13:17, 5 November 2025 (UTC)reply

A script is being coded to solve this issue phab:T409323. Yug (talk) 17:16, 5 November 2025 (UTC)reply

Migrating_`Translations:*`_pages

Hello APatro, hello Martin_Urbanec,

I'm currently leading the migration of WM's Lingualibre.org to Commons, approved by both LL and WC communities, and your expertise may be helpful.

I just imported a part of lingualibre wikipages to Commons this past week. I'm now investigating how to export/import the Translations:* pages and would like your guidance on this. I see two avenues forward.

(A) Via double server access : User:APatro (WMF)/Move translatable bundles across wiki which you (APatro) authored shows a way forward. I don't have server access today, but I could

get server access on lingualibre.org,
use your export script for my (~28) pages with translations and create as many xml dumps,
publish those xml files on github;
share with someone (Martin?) with WMC server access who would use the script to move those Translations:* pages into Commons

(B) Via xml Special:Export / Special:Import :

I inventoried the wikipages (~28) with translations (~5639) to migrate
Use :lingualibre:Special:Export to get their root xml files
Use Commons:Special:Import to import those (~28) root xml files
request a WMC translation admin to manually activate the translate module on those (~28) pages
Commons:Special:Import for those (~5639) translation subpages. Most translation pages are tiny and that will be split so that should work too.

Since I don't have any of the two server access needed for (A), I currently favor avenue (B). But I would also like to hear from you :

With avenue (A) and your 4-steps plan, do you know someone with Commons server access to lead the import on Commons ?
With avenue (B), do you see flaw in that 5 steps plan ?

Hope you may inform me a bit on these two approaches. Wish you a good day and thank for your tools. Yug (talk) 13:34, 2 January 2026 (UTC)reply

For (A), you can file a Phabricator task tagged with phab:tag/wikimedia-maintenance-script-run, and someone with server access will pick it up. You can also attach the XML export directly to that task instead of uploading to GitHub. —Tacsipacsi (talk) 00:23, 4 January 2026 (UTC)reply

+1 to @Tacsipacsi, except the attachment probably would not work (Phab has very low storage limits). Also see https://commons.wikimedia.org/wiki/User_talk:Martin_Urbanec#c-Martin_Urbanec-20260106143200-Yug-20260103192900 on my talk page for details. Happy to help with this once the ticket exists. Best, Martin Urbanec (talk) 14:33, 6 January 2026 (UTC)reply

@Tacsipacsi, @Martin Urbanec, @APatro, @Michael Barbereau WMFr hello,

Correct me if I'm wrong since I'm unfamiliar with Gerit search and Mediwiki developments. But I just noticed Lingualibre.org uses MediaWiki 1.35.2 (28 May 2021) while Patro's export script was edited in the [MediaWiki Language Extension Bundle](https://www.mediawiki.org/wiki/MediaWiki_Language_Extension_Bundle)'s gerit in 2023.

From what I understand, a mediawiki 1.35.2 won't have this ExportTranslatableBundleMaintenanceScript.php unless we update the MLEB to its v.2024.04 (?) which requires... MW 1.40.0. So, it looks like an impasse (?) Yug (talk) 16:08, 6 January 2026 (UTC)reply

The functionality seems to be added in https://www.mediawiki.org/wiki/MediaWiki_Language_Extension_Bundle/2023.07 which requires MediaWiki >= 1.39.0

The script wouldn't work with MediaWiki 1.35.2 APatro (WMF) (talk) 16:44, 24 February 2026 (UTC)reply

Pages to mark for translation

Done. Yug (talk) 20:33, 31 May 2026 (UTC)reply

Hello Vigneron, I come back to request your help as translation admin. Is it possible to mark the following pages for translations, either now (before creation) or when they become blue :

Ceci est à remarquer, mes excuses :

Best regards, Yug (talk) 18:03, 6 January 2026 (UTC)reply

@Yug: we don't use the translation extension on Category and Template (they already have their own systems, which you will probably import next). I'll wait for the other pages. Cdlt, VIGNERON (talk) 18:24, 6 January 2026 (UTC)reply

Oh, ok, received. Thanks for the reminder.

Commons:Lingua Libre/Misleading items import worked. All good. Yug (talk) 20:49, 6 January 2026 (UTC)reply

C'est bien ça : le « marked for translation» est remis à zéro quand je réimporte la racine de page. Il faut que j'enlève les racines de mes seconds imports avec les sous-pages. Je viens de faire un import des Helps pour rien : les racines sont repassées en non-traduisible, donc les imports des sous pages sont rejetés. Je recule sur ce coup mais j'ai compris. Yug (talk) 15:27, 7 January 2026 (UTC)reply

Audio files problem

Done

I uploaded some audio files for Kyrgyz words, but Wikimedia considers them not audio Almanbet Janışev (talk) 08:52, 8 January 2026 (UTC)reply

Please see Commons:Village pump/Technical#WAV files being uploaded with wrong MIME type. Thanks. Tvpuppy (talk) 12:24, 8 January 2026 (UTC)reply

@Almanbet Janışev It appears the problem has now been fixed. Thanks. Tvpuppy (talk) 21:03, 8 January 2026 (UTC)reply

[News] Are there broken wikipages in the project ?

Hello everyone,

In early January 2026 nearly all wikipages from Lingualibre.org/wiki/ have been migrated :

Help:Lingua Libre (subpages) for help pages
Commons:Lingua Libre (subpages)for project related pages
- Commons:Lingua Libre/List (subpages) for vocabulary lists

See also the Lingua Libre inventory tool. The tool filters out obvious obsolete page, fulther hand filtering was done. Click on the icon to see what have been migrated (✅), and did not (❌) in the last phase.

This migration implies contents changes, by examples :

Obsolete pages (❌): we did not migrate LinguaLibre:Administrators'_noticeboard → Commons:Lingua Libre/Administrators'_noticeboard, references to it are now misleading and can be removed.
- Some pages may need to be archived or even delete.
Outdated pages: some pages may need to be rewritten
Minor fixes: wiki links, categories, templates and code may be broken and require a quick fix. Some templates being different on Lingualibre.org and Commons, then now may be confusing and to remove.
Lingua Libre Qids: may be present in discussions but Qid links broken.

If you see such broken content, please jump in and fix it whenever possible. If not, please report the issue here.

As for translated pages, help is most welcome on the English versions first. We will later call for translators to help update minorities languages translations. Yug (talk) 12:30, 12 January 2026 (UTC)reply

Semi-protection

@VIGNERON hello, I would need your help to semi-protect those pages :

Those few pages are embedded in the soon to be deployed Lingualibre.org Yug (talk) 14:24, 27 January 2026 (UTC)reply

Done Cheers, VIGNERON (talk) 18:08, 29 January 2026 (UTC)reply

Thank Vigneron 👍🏼. Yug (talk) 19:15, 8 February 2026 (UTC)reply

Help:Lingua_Libre/de is not using german

Done. Fixed by Nortix08 and myself. Yug (talk) 20:33, 31 May 2026 (UTC)reply

None of the help pages display content in German. Nortix08 (talk) 11:25, 28 January 2026 (UTC)reply

Hello @Nortix08. Any idea what language is it ? You may fix this German page by editing the translations Yug (talk) 12:24, 28 January 2026 (UTC)reply

@Yug I've translated the messages in August 2021 into DE. In June 2025 they were changed. Is there a way to restore the old DE translations while retaining the LTZ(LU) translations? Nortix08 (talk) 16:19, 28 January 2026 (UTC)reply

Oh, can you check again ? I just noticed the help pages were not marked for translation, so maybe the German versions were hidden. I marked them. Yug (talk) 13:12, 28 January 2026 (UTC)reply

@Yug Cleared the Cache, forced a reload: Still not German. It's Letzeburgesch or Luxembourgish (ISO 639-2 "ltz") Nortix08 (talk) 16:10, 28 January 2026 (UTC)reply

Thanks for the clarification @Nortix08, I think a previous users incorrectly used the German page to input Luxembourgish (ltz). So I just copied those translations to the correct page (/lb), and you may now fix the German translations. You know how to do so right ? It's just a click + translate a few words.

Let me know if other pages have the same issue. Yug (talk) 19:59, 28 January 2026 (UTC)reply

Thanks @Yug and done. Nortix08 (talk) 08:36, 29 January 2026 (UTC)reply

Redesign

@Yug, is it okay if i redesign lingua libre project pages on commons? the changes will primarily be about using more codex stuff :3 əkrəm. 19:36, 30 January 2026 (UTC)reply

Hello @Əkrəm. Which page(s) are your thinking about ? Could you create and work on parallel drafts ?

(The following pages are embeded into https://Lingualibre.org and are therefore more sensitive :Commons:Lingua_Libre/Privacy_policy ; Commons:Lingua_Libre/About ; Help:Lingua_Libre.)

Yug (talk) 20:24, 30 January 2026 (UTC)reply

I was initially thinking of Commons:Lingua Libre, Commons:Lingua Libre/About and Help:Lingua Libre, then some other subpages as well if necessary. yeah, i will create previews in under my user page before deploying them to the actual pages, nw ^^ əkrəm. 08:50, 31 January 2026 (UTC)reply

Hello again,

Commons:Lingua Libre is a front mostly composed of /news/* subpages. These subpages will soon be phased out, so no need to refresh those.

The following pages are also imported into the future website, so sticking to Wikimedia Codex and Tailwind css would help :

I occasionally work on those as well to make them use Tailwind's layout imported via . Yug talk) 16:00, 31 January 2026 (UTC) Hello @Əkrəm,reply

Some user requested that Commons:Lingua Libre/List be more visible. I added the content, but I lack of ideas for beautifying it. Any idea ? Some simple colored box ? I did this in the past but it had more content. Let me know if you want to explore redesigning this List page. Yug (talk) 16:41, 30 May 2026 (UTC)reply

@Yug, sorry for the belated reply & previous inactivity. can you check Special:PermaLink/1223618794 and let me know in case you have any feedback? thanks in advance! əkrəm. 18:32, 31 May 2026 (UTC)reply

@Əkrəm, it's absolutely AWESOME ! Thank so much !

Make `Search` and `Create` each width:50%; but it's an easy fix.

I let you save your version upon the page, it's clearly better. Yug (talk) 19:21, 31 May 2026 (UTC)reply

glad to hear that, @Yug, thanks much! about the 50% thing, the flex: 1 that comes before padding: 0px 12px 12px 12px should handle it, but long story short, mw:Extension:InputBox handles styling of different kinds of input boxes a bit differently, therefore flex: 1 doesn't have any effect in this case. i'll use a temporary css fix for now, until the extension gets updated (if it ever does). thanks again! əkrəm. 20:15, 31 May 2026 (UTC)reply

it should be

done, please let me know if further adjustment is necessary! ^^ əkrəm. 20:55, 31 May 2026 (UTC)reply

Hello Əkrəm, Just letting you know, i m also continuing to test with the page's text content. Mostly: show subcategories so the users get a sense of how many lists we have. So the list of categories may temporarily get a bit messy. I will progress on this, then maybe roll back my edits. The wiki way. Yug (talk) 05:01, 3 June 2026 (UTC)reply

Only_one_linked_user_allowed_on_lingualibre

Message sent to User_talk:Richard_Nevell_(WMUK)#Only_one_linked_user_allowed_on_lingualibre. Yug (talk) 17:16, 3 February 2026 (UTC)reply

Lingua Libre mqg metadata

Done. Yug (talk) 20:33, 31 May 2026 (UTC)reply

Hello @ERHAMSYAHKDIKUTAIADATLAWAS & @Herpking,

I noticed your edits on mqg Lingua Libre records. Just letting you know that I edited the following items on Lingua Libre.

Records :

Speaker :

Erham (Q1573894)

I updated their metadata to :

fix their P3 recording URL pointer values : File:LL-Q1573881_(mis)-*.wav → File:LL-Q12952778_(mqg)-*.wav
fix their P4 values : Q1573881 (now deleted, was a copy of) → Q1573734 which represents Wikidata d:Q12952778, en: Kota Bangun Kutai Malay / id:Bahasa Kutai Kota Bangun.

These edits on Lingua Libre.org align and catch up with your filename changes on Wikimedia Commons. Yug (talk) 17:02, 3 February 2026 (UTC)reply

Adding new languages to record

Done. Yug (talk) 20:33, 31 May 2026 (UTC)reply

Hi, I can't add new languages to Lingua Libre. Can you please help me by creating the following so we can record some words and expressions? We are conducting a sound recording lab for language revitalization in Peru.

Kakataibo, ISO 693-3: cbr, Wikidata ID: Q5359560
Matsés, ISO 693-3: mcf, Wikidata ID: Q2981620
Awajún, ISO 693-3: agr, Wikidata ID: Q1526530
Ashaninka, ISO 693-3: cni, Wikidata ID: Q3437230
Kapanawa, ISO 693-3: kaq, Wikidata ID: Q2937196

Thanks you in advance, --Felino Volador (talk) 15:04, 6 February 2026 (UTC)reply

I will add them today. Yug (talk) 18:24, 6 February 2026 (UTC)reply

Done, Felino Volador, thank for your request. Yug (talk) 18:55, 6 February 2026 (UTC)reply

@Yug fantastic, thank you very much. Felino Volador (talk) 01:21, 8 February 2026 (UTC)reply

Q25337833 (mis) into Q7259896 (puj)

Done - (1) 🦊 Rename Or Replace (RoR) to rename files ; (2) RoR again to edit the wikitext ; (3) Mediawiki:Gadget-ACDC.js to update the SDC P:P407 `language of work`, see also my User:Yug/common.js) ACDC config, for a remove + add = update action. Yug (talk) 14:42, 14 February 2026 (UTC)reply

Hello, I think the 500+ files currently in Category:Lingua Libre pronunciation-other (Q25337833) should belong in Category:Lingua Libre pronunciation-puj instead.

It seems to me that these two wikidata item refers to the same language:

wikidata:Q25337833 (bahasa Punan Tubu) - Category:Lingua Libre pronunciation-other (Q25337833)
wikidata:Q7259896 (Punan Tubu) - Category:Lingua Libre pronunciation-puj

If this is indeed the case, can you please rename these files and move them to the correct category? Thanks. Tvpuppy (talk) 09:30, 9 February 2026 (UTC)reply

Hello Tvpuppy,

I spent the day developing a mediawiki tool, ReplaceOrRename.js, for this kind of situation. So we will be able to fix that soon.

But we want to go slow here : you pointed out there are two wikidata items involved, why ? If those are the same language, shouldn't the wikidata items be merged ? (or one delete) Yug (talk) 00:37, 11 February 2026 (UTC)reply

Ok I see. A bit late here, let's do it tomorrow: (1) renaming on Commons ; (2) updating the languages list on the app (remove Q25337833). Yug (talk) 00:51, 11 February 2026 (UTC)reply

Hello @Yug, thanks for the reply and your development work.

Apologies for not being clear in my message before, I was planning to merge these two wikidata items, but I guess I just want someone to confirm these two items are indeed the same before I perform the merge, since I am not familiar with this language.

Anyways, I have now went ahead and performed the merge in Wikidata. Thank you and have a good night. Tvpuppy (talk) 01:07, 11 February 2026 (UTC)reply

Hello again Tvpuppy,

With the wikidata merge, all files are now in Category:Lingua Libre pronunciation-puj, but their filenames and SDC still refers to Q25337833 (mis). I have one userscript and one under coding to fix that.

It looks ready, but could you double check somehow before I/we implement those userscripts corrections and fully merge those datasets ? The voices I can identify are on the files bellow.

Yug (talk) 10:24, 11 February 2026 (UTC)reply

Hello @Yug. I'm not sure exactly what to check, but I just did a brief look on the files below and they don't seem to have any other problems, so I think you can go ahead and implement your userscript to rename these files. Thank you very much for your help. Tvpuppy (talk) 17:29, 11 February 2026 (UTC)reply

Hello Tvpuppy,

Done. Files renaming have been successful using 🦊 Rename Or Replace (RoR). Filenames are fixed, but SDC data is still obsolete. I'm working on a tool to mass fix SDC as well well but it wont be ready rightaway. Yug (talk) 19:36, 11 February 2026 (UTC)reply

@Yug I see, thank you very much for the renaming. Tvpuppy (talk) 19:52, 11 February 2026 (UTC)reply

LL-Q7259896_(puj)-Baun (Yulius Nasarius)-a' la'ung.wav
LL-Q7259896_(puj)-ALFONSIUS26-anpai lun nuh.wav
LL-Q7259896_(puj)-Yukang (Julkarnai)-bibik.wav
LL-Q7259896_(puj)-Julkarnai-kah.wav

SDC fix

See also Help:Lingua Libre/Corrections.

@VIGNERON, @Sukkoria, @Lucas_Werkmeister hello,
About 450 files uploaded under an incorrect languages Qid got their filenames fixed. But their SDC also need to be fixed. Are you aware of an existing userscript, so for a given category or list of pagesnames, and some Property/Value & Property/qualifier/value directives, it update their SDC such as we could change:

P407 (language of work): Q25337833 → Q7259896
P9533 (Transcription) > Qualifier: P407 = mis → puj

I have a script under work, but if something good already exist I'm interested to use existing tools. Note: Help:Gadget-ACDC adds claims but i believe it does not update (remove + add) claims. Yug (talk) 20:25, 11 February 2026 (UTC)reply

There is several tools (indeed mostly for adding, see the list on Commons:Structured data/Navbox). I'll look to do it with OpenRefine this weekend. Cheers, VIGNERON (talk) 20:33, 11 February 2026 (UTC)reply

Lucas gave me a more detailed indication of AC/DC's capabilities, I have to request the claim's deletion then its creation with the correct value. VIGNERON, I will also explore your source. With all this I should find my way. Yug (talk) 20:44, 11 February 2026 (UTC)reply

Done :

Modified object	Action	Before	After	Tool
Filename	Replace : update the language	`LL-Q25337833 (mis)-Username-house.wav`	→ `LL-Q7259896 (puj)-Username-house.wav`	🦊 Rename Or Replace (RoR)
Wikitext	Replace : update the language	`\| language = Q25337833`	→ `\| language = Q7259896`	🦊 Rename Or Replace (RoR)
SDC P:P407 `language of work`	1. Remove, 2. Add : update the language.	P407 =`Q25337833`	→ P407 = `Q7259896`	Mediawiki:Gadget-ACDC.js, see also my User:Yug/common.js) ACDC config
SDC > P9533 transcription > qualifier P407 language of work		`mis`	→ `puj`	I don't know how to mass edit the qualifier while preserving the transcription value.

cc User:VIGNERON for the last item. -- Yug (talk) 14:45, 14 February 2026 (UTC)reply

Bug(s) report by {Mathsou}

Done Yug (talk) 20:33, 31 May 2026 (UTC)reply

Bug name: No visible fields No visible fields
Place: "Pick your profile
Incorrect behavior: Impossible to see the items to select for gender and licence + in a second try, impossible to even see the mentions "gender" and "licence"
Wanted behavior: Being able to see the fields "gender" and "licence" + being able to see and browse the list of elements + being able to choose an element from the list
Device: Laptop DELL
Web browser: Google Chrome
Phabricator task: No idea what this means...

Mathsou (talk) 10:52, 2 March 2026 (UTC)reply

I checked and I got the same bug as well. I noticed in the code, the select box for language proficiency (line 131), which is working, uses cdx-select. However, the select boxes for gender and license (line 172 and 182), which are not working, uses CdxSelect. Perhaps this is not related, but it is just a difference I noticed in the code. Thanks. Tvpuppy (talk) 12:29, 2 March 2026 (UTC)reply

Hello Mathsou and Tvpuppy,

I assume you are using lingua-libre.toolforge.org, this site is our test website. It's experimental and unstable yes. Yesterday tests were intense: we've been testing data structure changes, so there have been about 3~6 hours with mismatch on gender and license while we where testing the new code and its effects. It now work again with the new code and optional field (gender). Yug (talk) 12:10, 4 March 2026 (UTC)reply

Yes, apologies I should have clarified that I was using the test site. I can see the problem is gone now, so thank you very much for the fix. Tvpuppy (talk) 12:15, 4 March 2026 (UTC)reply

It's all good, happy users are enjoying the demo. Thumbs up to you for exploring the repository ! Yug (talk) 18:44, 4 March 2026 (UTC)reply

Unblock servers

Done, after assessing options, Michael and myself moved the whome project to https:/lingualibre.wmcloud.org. Yug (talk) 20:33, 31 May 2026 (UTC)reply

See also meta:User_talk:AmandaNP#Lingua Libre, User:57.128.108.213 and User:54.38.224.175, Blocks and protections#Unblock WMFR's servers.

Hello @VIGNERON, we may need your help again. The IP addresses of our both test and production servers are under global blocks. After digging and requests on meta, there seems to be a possibility Commons administrators like yourself may unblock or add a local IP exemption to our IPs, User:57.128.108.213 and User:54.38.224.175. Could you explore this possibilities and let me know if how it goes ? Yug (talk) 18:26, 4 March 2026 (UTC)reply

Hello @VIGNERON, as per the links in this section's top, we are requesting local WMC unblock of WMFR's servers, while we will plan for WMCloud hosting on the longer run. Yug (talk) 12:42, 5 March 2026 (UTC)reply

@Yug: I see that you wrote a formal request on Commons:Administrators'_noticeboard/Blocks_and_protections. Please wait and see. Cheers, VIGNERON (talk) 13:00, 5 March 2026 (UTC)reply

The request above was ignored and archived by admin. I therefore set WMCloud as my primary approach. Yug (talk) 14:00, 13 March 2026 (UTC)reply

We have now moved to WMCloud.org with our first Hello world there, which bypass the OVH IPs range block above. Yug (talk) 17:24, 18 March 2026 (UTC)reply

SPARQL endpoint status

Hello, Is the SPARQL endpoint https://lingualibre.org/sparql still maintained? Are there plans to revive it somewhere?

Thank you! Anton Kokh (WMDE) (talk) 10:03, 6 March 2026 (UTC)reply

Hello Anton Kokh (WMDE),

After the deployment of the new app the Sparql endpoint won't receive any new data.

The Sparql endpoint will still be available for about a year or so.

The complete and up to date data will be available via https://query.commons.wikimedia.org (not really available via Open API) or via the WM Commons Action API.

We would like funding to build a dedicated open API with mirror data at lingualibre.org but this is just an idea at the moment. Yug (talk) 16:05, 6 March 2026 (UTC)reply

Hello Anton Kokh (WMDE),

What usage you have in mind ? I'm not aware of any such usage by WMDE. It would help you if I can add that usage to my mind map since I'm de facto piloting Lingua Libre at the moment. Yug (talk) 13:59, 13 March 2026 (UTC)reply

Hi @Yug

The SPARQL endpoint is on the Wikidata allowlist, and we're doing some spring cleaning there - removing the endpoints that are consistently non-operational. Anton Kokh (WMDE) (talk) 08:21, 17 March 2026 (UTC)reply

I see. As of now it still stands. Phasing out date is not decided yet, and not before 2027. Yug (talk) 17:08, 18 March 2026 (UTC)reply

Sorry, just to clarify:

You mentioned that the SPARQL endpoint will be available for a year or so until it phased out.

But when I try making any kind of SPRARQL queries to it, it always fails. It seems like it has been down for months.

It is a known problem, or am I misunderstanding something? Anton Kokh (WMDE) (talk) 15:14, 4 May 2026 (UTC)reply

Hello Anton Kokh (WMDE),

I got similar failure. This Sparql tech is tricky to maintain online and occasionally crash. We rely on user's reports to reboot it. cc User:Michael Barbereau WMFr for this one. Please note that we are in th middle of a major Lingua Libre update (see Commons:Lingua_Libre#Announcement_(5)_:_Planned_release_!), so I have no idea when this SPARQL endpoint will get its reboot. But yes, this endpoint will be online for 2026 and 2027. Yug (talk) 14:34, 10 May 2026 (UTC)reply

WMC API & Tatoeba.org

Other example of API searches are shown in Help:Lingua Libre/APIs.

Hello Tatoeba,

Thank for contacting me via messaging. As for your question, there are always various ways to search Wikimedia Commons via API, a few classic ones :

haswbstatement:P11760%20intitle:aniversari — files with properties P11760, with "aniversari" in the title.
haswbstatement:P407=Q7026 — files with language of work (P407) being Catalan (d:Q7026).
haswbstatement:P407=Q150%20intitle:eau%20intitle:.wav — files with language (P407) being French (Q150), containing in their filenames `eau` and `.wav`.

For each such query, the API Result Limits are : srlimit=500 items per query for users, srlimit=5000 items per query for users Bots/Admins.

Each user should query WMC from their personal IP. The Reading Rate Limits are as follow :

Anonymous: ~1,000 requests per hour (IP-based).
Authenticated: ~10,000 (User-based).
Requirement: Must provide a unique User-Agent header.

I hope those informations can help you to sail smoothly. Yug (talk) 18:26, 16 March 2026 (UTC)reply

[EDIT]haswbstatement relies on a wikimedia Commons specific mw:Help:Extension:WikibaseCirrusSearch plugin, which does NOT support search with qualifier. It is therefore impossible to search by Recordist (d:P10893) or Spoken by (d:P10894) as visible in our new files, ex : File:LL-Q7259896_(puj)-Baun_(Yulius_Nasarius)-a'_la'ung.wav#P10893.

Source: mw:Help_talk:Extension:WikibaseCirrusSearch#P_=_some_value? Yug (talk) 19:44, 7 April 2026 (UTC)reply

Looking forward to recording https://tatoeba.org/en/sentences/search?from=ldn&sort=words&sort_reverse=yes&trans_filter=limit ! — Arlo James Barnes 22:14, 12 May 2026 (UTC)reply

[Bot] Regarding Lingua Libre Bot

Displaced from wd:User_talk:Yug#Regarding Lingua Libre Bot.

Hello, I have noticed that the bot has not been operating on Wikidata since mid September, and was advised to ask you about it. Is the bot being decommissioned or something? Or can we expect it to return someday? Redmin (talk) 09:20, 17 March 2026 (UTC)reply

Hello Redmin,

The bot is inactive since Sept because something on the server failed. It awaits to be restarted, but it has no active botmaster for the past year or so.

Also, as we are on the way to new recording app, the bot will need a coding revamp to move its queries away from the current Sparql & lingualibre.org source to then rely on Wikikmedia Commons API and haswbstatement queries. This isnt difficult but still, it needs a botmaster to code it, test it, and redeploy it, which we currently dont have.

Overall, User:Lingua Libre Bot is indeed going through a tough year but it shall reborn sooner or later. Yug (talk) 20:52, 17 March 2026 (UTC)reply

Oh, that’s unfortunate; thanks for your efforts though! Redmin (talk) 00:07, 18 March 2026 (UTC)reply

About custom categories

Done. See also #Adding categories with 🦊 Rename Or Replace (RoR).

Greetings all. I am preparing the interphase for the Wiki Audio Walk campaign which we started last year. But my major challenge had been customizing the Category under which the recordings done in a particular dialect which has no ISO code can be given a special upload funnel and all recordings done in that dialect labelled in a customized label.

I don't know if Lingua Libre now has a provision for that. Warm regards, Kambai Akau (talk) 22:47, 20 March 2026 (UTC)reply

Hello Kambai Akau, happy to read your message.

I noticed your work on those project and categories, so i ve been testing avenues to mass edit Lingua libre files. As of now, I would suggest —if you or a member of your team is familiar with regex— to use 🦊 Rename Or Replace (RoR)'s replace function. To add a project category, you may match the end of the wikitext for match /\}\}\S*$/ to replace by a variation of }}\n[[Category:Wiki Audio Walk-Tagalog]] (test needed to confirm)

I simpler alternative is as follow: I would like to provide the user, within the step 6 « Reviews / Publish », a typeahead-search field for custom categories. But its just in my personal wishlist at the moment. Yug (talk) 20:12, 22 March 2026 (UTC)reply

Thanks, @Yug! This is helpful. Neither I nor any member of my team has used Regex before now, but I will try it and see. I will try the second method as well. I think one or both of them should work! When I confirm, I will let you know. Kambai Akau (talk) 02:06, 27 March 2026 (UTC)reply

Hello Kambai Akau. Hey, your question rose again in #Adding categories , I indeed solved it with 🦊 Rename Or Replace (RoR). It works quite well. Just go slowly : it's a mass edit tool. Yug (talk) 20:33, 31 May 2026 (UTC)reply

Concerning Petscan and ML voices

Done. Fixed by myself. Note: This conversation is a shortened copy from an email exchanges, with consent of both parties. Yug (talk) 20:33, 31 May 2026 (UTC)reply

Hi Yug,

As you are familiar with Lingua Libre, this might be a great chance for me to finally get my questions answered - very few people got into Lingua Libre, indeed, so far I have been trying in vain to find assistance.

My first question is technical, i create lists using Petscan

https://petscan.wmcloud.org/?psid=18281491

I would like to get them into Lingua Libre so that a native speaker just has to read the words (here: names of female Court Presidents in Germany). Categories should be given once and then transferred to all the items of the list. Is this possible?

My second question is: Don't I contribute by generating files with Lingua Libre to fake voice messages generated by AI using my (and other) material? Reisen8 (talk) 11:16, 22 March 2026 (UTC)reply

Hello Reisen8,

Thanks for contacting me about Lingua Libre, i m always happy to discuss its usages, reuses or other concerns. As for your two questions:

In the new lingualibre app, "List" step, we have an "External tools" button on the right side. This allows you to provide a Petscan link. I patched the code to remove the `_` (underscore), it should work for you when you try.
Note : Since Petscan has changed in recent years, LinguaLibre.org legacy app's support for Petscan has been broken. Only the demo app, more recently coded, supports this new Petscan.
Your records will be freely available to everyone to use, yes. Reuse of our Wikimedia Commons' recordings for ML is one possible reuse. But this should be moderated as well, our audio datasets are not ideal either for such a project, not large enough —requirement is generally about 1000s of sentences recordings—, and your personal records will only be one voice among ~3000 ones.

Best regards. Yug (talk)

The patch is live. It will work : https://lingua-libre.toolforge.org Yug (talk) 11:25, 22 March 2026 (UTC)reply

[From email, with authorisation]
Hi again Yug,
I finally succeeded in completing the process you described, so now I can move on.
Another question:
Is there a way to get the sound files into the respective articles just behind the lemma in - let's say - the German wikipedia without doing that manually, one by one? That would save so much time.
Thanks again for your expertise ! :-) Reisen8 (talk)
Have a good week!

Hello Reisen8,

For this we need scripts and bots. Our User:Lingua Libre Bot is currently off and need a recoding to feet into the new processes. I raised this issue several times overs the years with Wikimedia France who manage funds from the French Ministry of Education. We may need to organize a kind of poll or wish page to get more light on this issue. Yug (talk) 16:39, 7 May 2026 (UTC)reply

Swadesh lists for Indonesia languages

I scrapped and imported ~2,300 Swadesh Lists into our Lingua libre lists. These provides a starter pack for some of the 50 indigenous languages you work with:

I created an userscript to augment those lists with a larger language of your choice, see Commons:Lingua_Libre/List/Ind/Swadesh110 where i appended English.

Once it this bilingual state it gives access to bilingual dictionary for true begginers :

https://dev.lingualibre.org/Dictionary/List/Ind/Swadesh110 (wait for it to load).

It opens a new way to use Lingua Libre to create within 2h a minimal dictionary (See: Commons:Lingua Libre/Dictionary). I used it for Oci/Jardin, it's not perfect but still, let me know if you see an usage for your workshops and community. Yug (talk) 11:02, 2 May 2026 (UTC)reply

Seems like these lists use IPA-based orthography, many of which will need to be fitted to the existing orthographic practices of the community. Swarabakti (talk) 11:58, 2 May 2026 (UTC)reply

Swarabakti hello. Thank for your review of these lists.

Knowing your communities (number of participants, digital friendliness, energies, mastery of their language's writing system), do you see an avenue to improve such 110 items Swadesh lists ?

If needed we can add English or Indonesian via script, this to help your contributors and to kick start bilingual dictionaries. A long requested feature by multiples indigenous communities who spent time on Lingua Libre's recording session. It may be interesting for User:Ardzun, yourself, WMID and the various WikiTutur, WikiKhata, ... projects to know about those lists and features. Yug (talk) 13:20, 2 May 2026 (UTC)reply

@Yug: Swadesh 110 is well within the range of wordlists we usually work with; it needs around two hours with a native speaker to collect a comparable number of items. These wordlists could perhaps be used by the fieldworkers to verify with the communities, rendering them in agreed-upon orthographic equivalents before recording.

Comparative Austronesian studies generally use Blust's list (as in ABVD), which is much better in terms of providing lexical items that can be lemmas in AN languages (as opposed to e.g. Swadesh "bark" and "there" that elicit sum-of-parts phrases like "kulit kayu" and "di sana" in Indonesian).

Btw the Indonesian list you linked also mixes Standard (some overly formal) and Colloquial (Jakarta) Indonesian terms; both are valid but we need two separate lists for them. Bilingual dictionaries in Indonesia almost always use Std. Indonesian as the glossing language in the published version, though several works specify that they used regional Malay/Indonesian varieties when documenting the languages (e.g. Visser's Kalamang dictionary which also has Papuan Malay glosses for its words and examples). Swarabakti (talk) 15:23, 2 May 2026 (UTC)reply

Anyway. I think the bilingual dictionary concept is very interesting and would be useful for our programs. We typically work from an Indonesian wordlist, eliciting the target language equivalents, then record, so the lists collected will have the same Indonesian glosses most of the time. Swarabakti (talk) 15:33, 2 May 2026 (UTC)reply

@User:Swarabakti, happy to met a felow field linguist familiar with lexical lists ! I also just realised the Indonesian languages maps were your maps. Thank you for those.

Ok, good news knowing that your community can work with Swadesh and Blust lists. Swadesh110 and 207 lists are now available to you, you may want to check there is no mixed Standards in the /Ind/Swadesh110 since it will be used to add the glosses to other Indonesian lists. (I tested, we could also have list & tri-lingual dict (ex: `{word} : {Indonesian} → {English}`) via augmentation from the bilingual source /List/Ind/Swadesh110. But I plan to change the code.) The normal behavior should be to add one glosse from one source per one run of the script, Indonesian for your group. (Or Malay or Papuan Malay if we have it an you share the iso so I add it to the script). You let me know when needed.

For Blush lists, I'm not sure they are CC-BY.4.0, but if you have 30 or more open license lists to add I have a another script to import those lists into Commons.

As for recording, please note: bilingual lists will only work with the next version at dev.Lingua Libre.org/app/ (currently in unstable BETA). I'm in support if needed but you seems pretty techfriendly already. Yug (talk) 17:44, 2 May 2026 (UTC)reply

@Yug: I'll edit the Indonesian Swadesh110 soon.

Also, ABVD says it is licensed under CC-BY 4.0 but idk if there are exceptions. Swarabakti (talk) 00:07, 3 May 2026 (UTC)reply

Oh, interesting: there is CC-BY-4.0 license data at lexibank/abvd and lexibank/abvdoceanic, in XML files. The README.md(s) suggest they focused on the 210 Blust items, so there seems to be a way forward to integrate those into Lingua Libre pipeline ò_ó ! Yug (talk) 11:10, 3 May 2026 (UTC)reply

I sent the ABDV team a message on Github. Yug (talk) 12:21, 3 May 2026 (UTC)reply

[Bot] Getting sound files into Wikipedia articles without doing that one by one manually?

Is there a way to get the sound files into the respective Wikipedia articles just behind the lemma in - let's say - the German wikipedia without doing that manually, one by one? That would save so much time. Reisen8 (talk) 15:18, 14 May 2026 (UTC)reply

Hello Reisen8,

You pointed out a major shortcoming from Lingua Libre's project. Our User:Lingua Libre Bot is not actively maintained. It also now required a full revamp to rely on Wikimedia Commons alone for its future targets. We've pointed this out for past 4 years.

This stalemate also cause a very severe beneficiary biaise. While 340+ linguistic communities contribute benevolantly,, about ~60% of our Wikimedia reuses were identified as benefiting the French readers alone. Overall, less than 10 wikitionaries really benefit from Lingua Libre records.

The general landscape is as follow : there are 15 identified fronts to support Lingua Libre. I carry nearly all those fronts as volunteer —which is obviously not sustainable— with rare freelance (5% of my work). I occassionally succeed in gathering volunteers or modest grants. The past decade also shown clearly that the heavier tasks ideally requires clear-headed WMF/French Ministry of Culture funds, the later being managed by WMFR.

Incoming priorities include renovations of Lingua Libre Bot, API consolidation (research), assets navigations (« Sonotheque »), datasets generation, the Language gallery and activity dashboard. There is also room for R&D if we play smart and forward : audio quality control, work offline, buffering, text-to-speech with meta:Wikispeech, etc. cc @User:Luc Hervier WMFr, @User:Rémy Gerbet WMFr @User:Lepticed7 . Yug (talk) 09:49, 19 May 2026 (UTC)reply

Hello Yug,

Thanks for showing me that big picture, I was not aware of it. I conclude that spreading the idea of LL in the German community would be more than welcome! And that uploading sound files is only a small part of the job (usually the one you can find people for, at least for once). Uploading only becomes really useful once the files are integrated in Wikipedia articles (that's at least what I create them for). This seems boring (compared to generating the files), but if many cooperate, it will have an effect. I suppose you would be delighted if we could find people here for cooperation on the way to get this automated? But for the time being, I try to appreciate that humans are irreplaceable at least for some jobs :-).

I suppose you will offer some kind of presentation in Paris, and I am looking forward to it.

Greetings --Reisen8 (talk) 08:37, 25 May 2026 (UTC)reply

Hello Reisen8,

Actually, I was too overworked this year with polishing the app, site, wikipages migration, sever deployment and all. So I did not submit a presentation to Paris but I will hold a stand for sure. I need rest tbh. But I'm happy of the work done ! Yug (talk) 19:24, 31 May 2026 (UTC)reply

Hello Yug,

I totally agree with your plan to rest a bit after your gigantic work - we need you in the future, so take care! I'll certainly show up at your stand, with gummi bears from WMDE :-)--Reisen8 (talk) 17:18, 1 June 2026 (UTC)reply

Adding Categories

Done. See also #About custom categories, 🦊 Rename Or Replace (RoR).

How can I add a certain category (like: German-language surnames) to all sound files I create on a certain day without doing it one by one? Reisen8 (talk) 15:27, 14 May 2026 (UTC)reply

help:gadget-HotCat#Making more than one category change may help, although I'm not sure about doing it by date uploaded. Arlo James Barnes 19:22, 14 May 2026 (UTC)reply

Hello again Reisen8,

To add category you may use 🦊 Rename Or Replace (RoR) with its "replace" function, replacing the last line of wikitext with itself + your category. It will requires the list of target filenames, a smart "end of text" regex match, and slow testing but it should work. To get the list of files by date : browse your upload logs, copy the files, and clean this up. Yug (talk) 09:55, 19 May 2026 (UTC)reply

I tested it successfully.

Open RoR

Action : Wikitext replace ;
Source : Manual list > List your targets.
Match : ^([\s\S]*)$
Replace with :

$1
[[Category:New Category]]

[x] Regular expression.

Yug (talk) 10:36, 19 May 2026 (UTC)reply

For categories, there is also Help:Gadget-Cat-a-lot, but I'm not really familiar with that tool. Yug (talk) 09:15, 21 May 2026 (UTC)reply

Liste de bugs ou de problèmes rencontrés

Done by myself, sorting points by category, fixed 7, implemented 5 wishes, answered the rest within half a day. Yug (talk) 20:33, 31 May 2026 (UTC)reply

Voici la liste des bugs et des problèmes que j’ai rencontrés :

Il n’est marqué nulle part où se trouve les listes locales d’enregistrements, ni comment les créer avec la nouvelle version.
Il n’est marqué nulle part où se trouve la liste des langues disponibles dans la base ni comment les administrateurs peuvent en ajouter de nouvelles.
Il n’est marqué nulle part où se trouve la base de donnée des enregistrements ni comment y accéder pour corriger les données.
La liste des partenaires ne liste pas tous les partenaires ayant participé historiquement à l’élaboration de l’outil.
Le menu « rechercher » n’explique pas que c’est pour rechercher parmi les enregistrements.
La section « see also » de la page d’aide est cassée (« v » ? « t » ? « e » ?).
La page « à propos » mentionne « l’onglet Record Wizard » alors que ce terme n’apparaît plus dans l’interface.
La page « à propos » parle des langues « signataires » au lieu des langues « signées »…
En fait, c’est tout le texte « Pourquoi participer ? » de la page « À propos » qui devrait être relu et réécrit dans sa version en français. Le style est lourd et plusieurs phrase sont incohérentes.
Mettre sa langue dans le menu en haut ne change pas les langues des pages autres que celle de l’enregistrement.
Le micro doit absolument être testé à chaque session, même si on enchaînes les sessions d’enregistrement (ça fait vraiment perdre du temps lorsqu’on enregistre à la suite plusieurs personnes).
Si on teste le micro avant de l’activer, le micro est testé quand même et on passe toutes les étapes avec un micro qui ne marche pas (ça sert vraiment à rien du coup…).
L’encart « LINGUA LIBRE DJANGO : VERSION 2026 » mentionne 10 000 langues sur Terre alors qu’il n’y en a que 7 000 (Franchement, c’est plutôt honteux de sortir des chiffres comme ça… Ça fait vraiment pas sérieux.).
Dans l’outil de sélection des mots :
1. On ne peut pas ajouter plusieurs listes locales, en choisir une efface tous les mots sélectionnés. Ça devrait seulement ajouter les mots de la liste de mots déjà sélectionnés, et pas effacer tout ce qui a été déjà listé.
2. L’outil « à proximité » ne marche que si la liste des mots à enregistrée est vide, il devrait ajouter les mots en plus de ceux qu’on liste déjà.
3. L’outil « catégorie Wikimédia » liste absolument toutes les wikipédias et tous les Wiktionnaires. C’est plus de 500 objets dans un menu liste, c’est illisible et incompréhensible pour le commun des mortels, de plus personne ne connait parfaitement le code ISO de sa langue. Pourquoi ne pas avoir gardé l’interface précédente qui était claire ?
4. L’outil « catégorie Wikimédia » ne marche pas (exemple avec : Avions en français pour fr.Wiktionary, mais testé avec une dizaine d’autres catégories). Il donne « Error fetching category members ».
5. L’outil « Outil externe » ne marche pas avec Petscan (exemple avec : [1]).
6. L’outil « Lexème Wikidata » demande un « modèle de recherche », ce qui ne veut rien dire… On n’a aucune idée de ce qui est attendu.
Dans les paramètres de l’outil d’enregistrement la phrase « Durée de silence signalan la cloture d'un enregistrement » comporte deux erreurs et doit être corrigée en : « Durée de silence signalant la clôture d'un enregistrement ».
Dans l’outil d’envoi des enregistrements, les enregistrements ne sont pas cochés par défaut, or, il est bien plus efficace et bien plus simple de décocher les enregistrements défectueux plutôt que d’avoir à tout cocher tout ceux qui sont bons. C’est d’autant plus important que la fonction « écouter tous les enregistrements » ne coche pas les enregistrements déjà écoutés.
Si on vérifie les logs de publication des enregistrements, on ne peut plus revenir sur la page précédente pour continuer sur lingua libre. Ces logs devraient être ouvert automatiquement sur un nouvel onglet. Pire, si on fait revenir en arrière avec le navigateur, on se retrouve sur la page de sélection des profils et elle ne marche plus, impossible de cliquer sur suivant malgré le profil sélectionné.
Il n’est pas possible de quitter l’application d’enregistrement et de revenir sur la page de l’outil en général.

Je tiens à signaler que je suis très déçu de la qualité de l’outil actuel. D’autant plus qu’il ne semble plus possible pour la communauté de s’emparer de la configuration (listes locales, listes des langues, données, etc.) et que l’on perd ainsi la résilience qu’avait l’ancienne version. Ce n’est plus un outil communautaire, on perd quelque chose de précieux. Lyokoï (talk) 00:56, 25 May 2026 (UTC)reply

Bonjour Lyokoi,

Être très déçu en ayant la chance de se positionner en juge sans autre effort depuis 3 ans sur ce projet que celui de commenter le travail que d'autres mènent avec des moyens très insuffisants... Enjoy ! Pour l'approche "on se retrousse les manches on code", c'est plus bas.

Si ton ton final me fatigue, je prends tout de même pour ce rapport de test et sa liste de souhaits. Comme annoncé dans le bandeau du site, nous espérons recevoir vos rapports de bugs. J'apprécie le travail accompli, et j'ai pris les bons éléments. Au moins 12 points et idées réparés ou implémentés. Nous avions une erreur dans le fichier nginx de la prod qui cassait les requêtes sur wikt et wmcloud, les bugs signalés les plus sérieux. Une bonne partie des autres points correspondent à des choix ou à un manque de connaissance du projet et des annonces de ces derniers mois. J'ai clarifié, réparé, ou implémenté tes différents points. Cela reste de l'open source largement maintenu via du bénévolat : il reste toujours des éléments améliorables sur lesquels tu peux contribuer.

Pour ton dernier paragraphe avec des accusations envers WMFR, l'équipe presta et bénévole (700h du code), je t'encourage à le supprimer vu que tu es à nouveau très mal informé malgré les 4 annonces officielles citant la migration des listes, éditables, sur Commons:Lingua Libre/fr, ainsi que la documentation de Aide:Lingua Libre/fr. Nos 6 000 (?) listes locales migrées sur Commons, la liste de langues (gitlab), et les données (Commons) sont toujours éditables en open source. Ton anglais est largement suffisant pour lire ces annonces, la dernière annonce était disponible en français.

WMFR n'avait pas le budget de sa politique, de très loin. On a fait avec 1/4 du budget nécessaire, ou 4 fois la charge du budget. A un moment ce serait bien de le reconnaitre, de reconnaitre le travail accomplis, et de reconnaitre le travail qui reste à faire.

Les 7 points 1, 2, 3, 4, 7, 8, 9 concernent des contenus rédigés par WMFR et/ou la communauté. Ils sont toujours documentés dans les pages d'aide. Aide:Lingua Libre sert maintenant d'entrée unique. Michael, Vigneron, Sukkoria et moi-même avons tous déjà relevé le besoin de mettre à jour ces contenus lors d'efforts communautaires futurs.

Info 1 « pas d'aide listes locales » : « Lisez le manuel ». Le point d'entrée unique est présentement l'onglet Aide (Help:Lingua Libre) > Créez vos propres listes. Il n'y a plus d'onglet principal "Listes" dédié pour le moment pour la raison suivante : la communauté possède 1) Help:Lingua Libre/Create your own lists, 2) Help:Lingua Libre/Lists, 3) Commons:Lingua Libre/List qu'il conviendrait de fusionner lors d'un éventuel éditathon. Lorsque la communauté aura fait ses choix, les dev pourront mettre la wikipage choisie en onglet principal, c'est désormais assez facile.
Info 2 « pas d'aide pour trouver les listes » : Idem, lisez le manuel. Le point d'entrée unique est présentement l'onglet Aide (Help:Lingua Libre) > Créez vos propres listes.
Info 3a « où se trouve la base de données » : Si tu parles des audios, idem, lisez la doc. Le point d'entrée unique est présentement l'onglet Aide (Help:Lingua Libre) > Télécharger l'ensemble des données documente l'accès aux données. C'est une page communautaire à mettre à jour. Dans la nouvelle topbar nous avons également l'onglet Datasets (module codé en bénévole), qui pointe vers des datasets plus ou moins à jour. Souhaites-tu une nouvelle feature quelque part ? Tu peux contacter WMFR, contribuer, ou me demander une feature que je coderai en bénévole. Pour la base de données utilisateur, il faut les accès admin et serveur.
Info 3b « comment accéder et nettoyer les données » : Pour nettoyer, un premier outil a été développé (🦊 Rename Or Replace (RoR), en bénévole) pour les titres de fichiers et le wikitexte sur Commons. Un second protocole est nécessaire pour corriger les SDC. User:VIGNERON avait une piste via OpenRefine. J'ai une autre piste.
Agree 4 « la liste des partenaires est à mettre à jour » : absolument d'accord. WMFR est au courant. Tu peux contacter WMFR ou contribuer.
Agree 7, 8, 9 « en fait c’est tout le texte [...] de la page À propos qui devrait être relu et réécrit » : Absolument d'accord. Idem précédent.

Les 3 points 12, 14.1 ; 14.2 sont normaux ou fonctionnent de mon côté :

Weak support 12 « si on n'a pas branché le microphone externe, l'étape est tout de même validée avec le micro interne » : oui, logique. Option : → Rédiger un meilleur guide pour l'étape 3. [EDIT]: On peut imaginer quelque chose.
Done 14.1 « charger plusieurs listes locales » : fonctionne de mon côté, prendre en compte la limite numéraire.
Done 14.2 « liste par proximité géographique » : fonctionne de mon côté.

Les 7 points (6, 10, 14.3, 14.4, 14.5 ; 15 ; 17) ont été pris en compte et sont déjà corrigés. (Une erreur Nginx s'était glissée lors de la migration).

Done 6 « see also ne supportait pas le modèle » : bien vu → réparé (module codé bénévolement)
Done 10 « changer la locale doit recharger le contenu wiki de la page en cours » : bonne idée → implémenté

Done 14.3 « catégorie Wikimédia » : choix, l'ancienne version était un texte ouvert, nécessitant l'ISO également, mais sans le proposer. Note : le sélecteur peut être amélioré.
Done 14.4 : nginx corrigé, fonctionne. Bien vu.
Done 14.5 : nginx corrigé, fonctionne. Idem.
Done 15 : traduction corrigée et live. Tu peux directement raffiner les traductions via translatewiki que j'ai mis en place.
Done 17 « ouvrir les logs dans un nouvel onglet ». Oui.

Cinq points (10, 13, 14.6, 16, 18) sont des préférences ou améliorations pouvant être raffinées par les bénévoles :

Info 5 « Menu Rechercher » : module développé en bénévole pour remplacer rapidement la sonothèque. Qui veut poursuivre peut poursuivre. Tu peux contacter WMFR ou contribuer.
Done 10 « changer la locale doit aussi recharger le contenu wiki de la page en cours » : bonne idée → implémenté.

Done 13 « 10 000 langues » : « mentionne 10 000 langues sur Terre alors qu’il n’y en a que 7 000 (Franchement, c’est plutôt honteux de sortir des chiffres comme ça… Ça fait vraiment pas sérieux.). » VIGNERON de son côté encourage à inclure « 20 000 langues suivant Wikidata », j'ai donc inclus les 7 000 bien connues et "un bonus Wikidata". Vous pouvez en discuter avec VIGNERON et me dire ce que vous préférez.
Done 14.6 « modèle de recherche » : J'ai modifié en "Schéma de recherche". Si tu trouves mieux tu peux raffiner sur translatewiki.

Weak oppose 16 « enregistrements décochés par défaut » : les audios sont volontairement non validés (non cochés) par défaut. Nous tenons à ce que le cochage soit un acte volontaire, que l'utilisateur soit responsabilisé quant à la qualité des publications. Nous avons tout de même ajouté les boutons "cocher tout" : l'utilisateur porte la responsabilité. Nous avons également un bouton "écouter tout" permettant au final d'avoir le fonctionnement rapide que tu décris. À moyen terme, c'est un secret de polichinelle, il faudrait insister davantage sur la qualité des audios (on voit passer des monstres).
Done 18 « retour page d'accueil ? » : L'icône conduit maintenant à la page de l'app.

J'ai implémenté les points pertinents, répondu aux autres, certains demeurent ouverts. Yug (talk) 13:27, 25 May 2026 (UTC)reply

D'autres bugs sont connus, suivis et en cours de résolution. On se remonte les manches avec très peu de moyen, 1000h de bénévolat ces 12 derniers mois, tous les devs sont présentement bénévoles. Yug (talk) 16:59, 25 May 2026 (UTC)reply

Username moving for Juwan

See also 'RenameOrReplace.js' cited above.

See also Special:ListFiles/NISENGWE3

requesting that the category and its associated items be moved from "JnpoJuwan" to "Juwan": Category:Lingua Libre pronunciation by JnpoJuwan. the username changed but the files did not, oddly enough. Juwan 🕊️🌈 14:47, 25 May 2026 (UTC)reply

Hello Juwan,

For now we can only do partial correction, I would not be able to fix structured data #P10893 and #P10894. Maybe we should gather those requests for now. Yug (talk) 15:01, 25 May 2026 (UTC)reply

Audio recordings for the Sanskrit language

I can no longer record the Sanskrit audio; there's an error I don't know what it is. Currently, I have to use Lingua Libre, which feels annoying and not as convenient to use as before. Lingua Libre also has a font error; it doesn't display the Mon text correctly. It's a mix of Zawgy font and Unicode font, that's why the Mon text aren't displayed correctly.𝓓𝓻.𝓘𝓷𝓽𝓸𝓫𝓮𝓼𝓪|𝒯𝒶𝓁𝓀 19:43, 25 May 2026 (UTC)reply

Hello 咽頭べさ, thanks for this feedback / bug reports.

I know the first bug in your current image which is tough to replicate.

I will need more context for your 2nd complain about font issues.

Could you share a screenshot as well ? Yug (talk) 19:50, 25 May 2026 (UTC)reply

@Yug Regarding the font error, I've explained it in this image. Please open it, thanks.--𝓓𝓻.𝓘𝓷𝓽𝓸𝓫𝓮𝓼𝓪|𝒯𝒶𝓁𝓀 20:02, 25 May 2026 (UTC)reply

咽頭べさ thanks again. Which one is correct, yellow or red ? Could you give us that same string of characters ? Yug (talk) 20:21, 25 May 2026 (UTC)reply

Hi @Yug, I've highlighted the font errors in red, and the ones in yellow are just examples. Try copying this Mon text from this "ဆာန်တှ်ေလဴ" The meaning is "if you love me, just tell me", thanks. 𝓓𝓻.𝓘𝓷𝓽𝓸𝓫𝓮𝓼𝓪|𝒯𝒶𝓁𝓀 05:02, 26 May 2026 (UTC)reply

404 for user of a submitted audio

Done by Michael WMFr. A bit more to do to properly style this new archive. Yug (talk) 20:33, 31 May 2026 (UTC)reply

On here https://en.wiktionary.org/wiki/File:LL-Q1860_(eng)-Flame,_not_lame-bitch-ass.wav this link is up: https://lingualibre.org/wiki/Q1191054 , but it 404's . ~2026-31929-09 (talk) 19:44, 28 May 2026 (UTC)reply

Thank for that one which we missed. I will create a redirect to https://archive.lingualibre.org/wiki/Q1191054 Yug (talk) 05:18, 30 May 2026 (UTC)reply

Done. Michael deployed the archives with redirection. Your issue is solved. We still have work to do on those archives. Yug (talk) 21:08, 30 May 2026 (UTC)reply

archive.lingualibre.org

Bonjour,

Qui peut modifier toutes les pages dans https://archive.lingualibre.org/ ?

Cordialement. —Eihel (talk) 16:44, 4 June 2026 (UTC)reply

Bonjour Eihel,

Les archives sont bloquées en écritures, elles sont mise en consultation seulement comme mentionné dans le bandeau de haut de site. Si tu souhaites en savoir davantage tu peux me contacter sur Discord en DM. Bien à toi. Yug (talk) 19:49, 4 June 2026 (UTC)reply

New userscripts for LL maintenance

See also Help:Lingua Libre/Tools and Help:Lingua Libre/FAQ.

CC User:Ardzun, User:Swarabakti, User:Rulwarih, User:Tofeiku.

Hello @Jesedmateo, Thanks for reaching out to me following your recording session with Wayuu speakers at the Encuentro de Comunidades Indígenas en los proyectos Wikimedia conference in, Riohacha, Colombia. May i encourage you to add the following tools to your meta:Special:MyPage/global.js page ?

/* ************* Lingua Libre toolbox *********************** */
// [[m:User:Yug/RenameOrReplace.js/intro]]
mw.loader.load( '//meta.wikimedia.org/w/index.php?title=User:Yug/RenameOrReplace.js&action=raw&ctype=text/javascript' );

// [[m:User:Yug/Lingualibre_upload_batches.js]]
mw.loader.load( '//meta.wikimedia.org/w/index.php?title=User:Yug/Lingualibre_upload_batches.js&action=raw&ctype=text/javascript' );

// [[m:User:Yug/Lingualibre_list_augmenter.js]]
mw.loader.load( '//meta.wikimedia.org/w/index.php?title=User:Yug/Lingualibre_list_augmenter.js&action=raw&ctype=text/javascript' );

These tools would help you with managing LL assets on Commons. We can discuss each one further when needed. Yug (talk) 05:25, 12 June 2026 (UTC)reply

Download datasets

See also Help:Lingua Libre > /Download datasets, /APIs, and /FAQ.

The text below is a gist of an email conversations between User:Yug and User:Locongres, shared due to content relevant to the Lingua Libre community. cc: User:Unuaiga at Locongres' request.

Hello @User:Locongres, Thanks for reaching out to me via email regarding to the Whistled occitan datasets.

Hi Yug. A while back you told me that the LinguaLibre's online datasets were outdated and that I needed to download directly them via Wikimedia Commons. Is there a simple way to do this that doesn't require hours of programming? I've already spent half an hour on it and managed to get the list of files in Category:Lingua Libre pronunciation by Univòc64. I tried downloading them via urllib, but it seems Wikimedia Commons is blocking it. How can we download the files? Thanks. --Locongres 19:27, 19 June 2026 (UTC)reply

Hello Locongres, happy to ear from you.

There is currently no easy access to the most recent records, we plan to revamp the broken datasets service but it will occurs during the next coding cycle (2026 Q3 & Q4). Given your email I looked into our previous scraper and I now also have the list of Whistled Occitan files. To get the files via their download URLs we have two methods : MD5 encoding of the filenames or API request with `image info`. (You may have tried either of them). But this is where it gets tricky... Around Autumn 2025, to respond to aggressive AI scrapping, the WMF reduced the leeway for bots and scrapers so they now get blocked faster. We now need to: respect ratelimits, use an XHR header identifying ourself, and ideally, use WMF-whitelisted servers and community-approved bot credentials (I have those). Q1: When do you need this script / dataset? (I have personal goals to finish before volunteering.)

Best, --Yug 09:26, 22 June 2026 (UTC)reply

Hello Yug, honestly I just want to download the whistled Occitan dataset. You told me that Lingua Libre datasets are only occasionally updated... So, which way should we go ? --Locongres 11:03, 22 June 2026 (UTC)reply

Hello Locongres, the legacy datasets system is down and the new datasets system isn't there yet. So I just ran an updated Python script from a whitelisted Wikimedia server to generate the zip file you need using some of the tricks mentioned in my previous message. It's a bit "raw" but the zip is now available here: https://lingua-libre.toolforge.org/datasets/oci-biu.zip. Best regards. —--Yug 11:06, 23 June 2026 (UTC)reply

SDC block uses wrong copyright property

when a pronunciation is uploaded to Commons and the SDC block is added, it assigns it the copyrighted, dedicated to the public domain by copyright holder (Q88088423) under copyright status (P6216) rather than copyrighted (Q50423863), despite my license being set to CC BY-SA 4.0. any idea what's going on/how to fix this? —aadenboy (talk|contribs) 05:58, 4 July 2026 (UTC)reply

Tracked in Phabricator
Task T433775

I noticed the same issue in the recent Lingua Libre recordings of TCraze. As I didn't find a related Phabricator ticket, I just created one. --Volvox (talk) 20:23, 1 August 2026 (UTC)reply

Please add Tyap dialects to Lingua Libre

Hi admin, please add the following Tyap [kcg] dialects to Lingua Libre.

Tyap Tyap Q3912765
1. Cacrak Tyia̱caat Q115138365
2. Fantswam Fantswam Q115138197
3. Gwolog Gwoot Q115138293
4. Sholyio Sholyia̱ Q115138052
5. Takad Ta̱kat Q115138342
6. Tuku Tuku Q115138406

Thanks! Kambai Akau (talk) 10:38, 23 July 2026 (UTC)reply

Pali Q36727 and Mon-Pali Q100271134

Both of these categories may be describing Mon-Pali pronunciation. I'm not sure if there are any changes needed to categorization.

Category:Lingua Libre pronunciation-pli (describes its scope as Mon-Pali, but is associated with Pali Q36727)
Category:Lingua Libre pronunciation-other (Q100271134) (Mon-Pali Q100271134)

— Wracking ( talk / contribs / uploads ) 19:04, 24 July 2026 (UTC)reply

Feedback from the Lingua Libre booth at Wikimania 2026

Following consent by WikiLucas00 this post was slightly edited by Yug to better categorize each point with tags such as [{type} - {workload}].

Hello, here is some feedback I collected, from a few days of having a dedicated Lingua Libre booth at Wikimania.

We used a PC with Ubuntu and Firefox to do the recordings, and created a dedicated Wikimedia account (User:LinguaLibreBooth), which means we had the same limitations as a new user (i.e. no specific user rights on Commons).

I want to be clear that this feedback (which is mostly issues) is not aimed at a specific person but at our community, to discuss similar experiences and workarounds, and list what we need to improve the project in the future.

Pinging a few users @Yug, Pamputt, Lepticed7, Lyokoï, VIGNERON, Poslovitch, Danÿa, Juwan, and Culex: , feel free to ping others as a response!

A) Speaker profile form

[Wishlist - easy] The field for the birth year is not required and should be more clearly optional. After discussing with Yug, the color of the star sign matters (it is green here), but maybe a "ⓘ" character with a tooltip explaining why this info can be useful would be more user-friendly.
[Wishlist - medium] The speaker profile form should ideally allow only some specific values, for instance when selecting a place of residence or the place where one learned a language, one should only be able to select locations, not other types of elements such as "Family name"

B) General UX of the Record Wizard

[Wishlist - deep] Some users complained about having to go trough all steps each time, when their setup does not change most of the time. Maybe a tradeoff could be found, but I know there have been some efforts on forcing the user to select the language before recording, to avoid mistakes.
1. Regarding that, I think we can agree that when the profile contains only 1 spoken language, the language picking step is useless and should be skipped.
[Wishlist - easy] When selecting a Wikimedia project as a source for the list of elements to pronounce, the language prefix should already be chosen based on the language selected one step earlier (with the option to change it manually if needed). In the same situation, when selecting a category in a Wikimedia project, the "Category:" prefix should already be present, having to type the language code and the category prefix is a huge friction point for users.
[Remark] Browsing local lists from the record wizard is not presented in the same way than before
[Question] I am unsure the bot of User:Olaf has been updated to maintain them on Commons.
[Wishlist - easy] I did not find the option we had in the past to shuffle the list, which is quite problematic as list order may impact the data

C) Recording

[Wishlist - deep] I noticed that users tend to not focus on the right panel showing the element they're currently pronouncing, but instead look on the left panel, containing the list of elements (which makes sense, as this allows them to prepare for the next word). With that observation, I also noticed this left panel does not autoscroll, once the user recorded the first x elements. Some users tried to scroll themselves, but this can disrupt their flow and generate some noise, so ideally the list should autoscroll to have the current element vertically centered in the left panel.
[Bug - deep] We still had the issue where some parts of the audio file are missing when hearing back the recording, resulting in an unusable file.
[Remark] The idea of adding the option to disable the Voice Activity Detection in the settings emerged from discussions with users. This would allow specific users to work at their own pace and to record longer texts without worrying about the silence threshold (no matter its value)

D) Uploading recordings

[Bug - medium] Very often when clicking upload, the first recordings fail, sometimes a big number of recordings fail to upload. That should not be immediately the user's problem, the tool should retry on its own and just be transparent that it's retrying, and if it encounters too many errors, it should stop and warn the user the uploading failed. There is no added value to ask the user if they want to retry uploading failed uploads.
[Remark] The worst thing that could happen is losing recorded data due to a problem on the client side (or by being blocked/limited by the Commons server and needing to close the browser). I think there should be an option to export the current recording session into a downloadable archive, and an option to upload such an archive into the record wizard, to continue an interrupted upload session.
[Bug - Commons limit ?] We had an issue several times, where uploads stopped after 60 files, no matter if we retried. Even worse, once the message "Recordings were uploaded successfully" appears, the checkboxes become grey and nothing can be clicked, resulting in lost data. We were far from hitting the 380 files/72 minutes ratelimit at that moment. Please see the attached screen recording.
Upload issue encountered during Wikimania 2026
[Wishlist - deep] Since we have a backend server, could we use it to handle the uploads a little better, as a background task, to avoid depending on the client staying open? This could also allow to upload a 1000-file batch at a rate which respects the 380 files/72 minutes ratelimit.
[Wishlist - deep] Regarding this ratelimit, could we try whitelisting LinguaLibre tags to allow all of our users to not be impacted by the ratelimit, without requesting the specific user rights?

E) Website and codebase

[Wishlist - medium] The search feature is nice but it does not work properly for now: it states searching for "exact word" but in reality it also retrieves elements containing the search term as a substring (e.g. searching for "café" in French also retrieves "pousse-café"). The now deleted LiLiZip Gadget was handling that better, it's probably just a regexp to update.
[Wishlist - easy] It's a good thing that the Gitlab repo is linked on the website, maybe we should also have Phabricator?
[Remark] Shouldn't the former Github repo be archived and point to the Gitlab in its README?

I'll let others who volunteered on the booth or who used Lingua Libre these last months add their feedback if they wish.

Thank you all for your contributions on Lingua Libre and let's continue the good work!

All the best — WikiLucas (🖋️) 11:39, 28 July 2026 (UTC)reply

Hello WikiLucas ,

Thank you for your forward enthusiasm on Lingua Libre, many ideas for further improvements and rapid bug reports. Allow me to respond in a semi-concise manner for now. Please also note that I reduced my coding contributions given past years heavy coding on this project, so these answers are purely informative.

Wishlist (10): You identified ~10 possible improvements which can be merged with current wishlists on Phabricator and meta. Most of your suggestions tagged `easy` are minor coding and just need someone to tackle them. A few other require medium or deeper coding and app-flow awareness, those could be addressed in future coding cycles supported by volunteers, hackathons or WMFR-managed grants.

C1 (When recording, users focus on the left list), this is known and this suggestion is actually a return to the former Shtooka UI design, which I also encourage. For PR (keep limited changes to limit complains) and for HR reasons (Developers Pushkar, myself, Aditya, Gilles already gave +1000h of free work these past 2 years), we left such radical change aside and for future coding cycles.
D4 (Backend buffer for uploads, bypassing ratelimit) have been discussed for years but represent an heavy workload.
D5 (Whitelisting Lingualibre) would be ideal, by it contrary to the user-by-user userright security approach on Wikimedia Commons, and is therefore unlikely to succeed.
E1 (Search/Sonotheque) : yes, fill free to fix the text or the regext. Please note this demo is a quick Sonotheque placeholder coded on my free time since the recoding of the sonotheque was not in the scope of past years missions. Could be part of future coding cycles, please ask WMFR.

« Bugs » (3): You reported 3 serious issues.

C2 (Part of the audio missing): Yes, this in on our radar but requires investigation. This depends on Lingualibre.js which was maintained by 0x010C and Wikimedia France. But maybe over reliance on Vuejs local memory is an issue. Requires a dedicated investigation. (Note: since we just migrated this block of code, this was likely occurring on the previous LL Wizard version).
D1 (First upload fails) : The nuisance have been fixed. Explainer: Lingua Libre is an living open source project. During the Indic hackathon, June 26-28, 2026, Pushkar mentored a junior dev. They consolidated Lingua Libre Steps 1: Locutor and Step 6: Upload. The Wikimania revealed this side effect which Pushkar7077 fixed a week later, see logs. We may thank him for this whole volunteer effort.
D3 (Limited to 60 uploads) : We investigated it and the most likely cause is Wikimedia Commons security policies. It only affected account registered less than 4 days before using LL, they seems limited to 60 uploads. Not much we can do, just anticipate and ask people to register 5 days before their major recording session.

Questions (1):

B4. Yes, in the scope of the List migration freelance (here), I contacted User:Olaf to ensure User:Olafbot continues to operate properly at the new list location at Commons:Lingua Libre/List/{Iso}/{List_title}. Olafbot's list also now leverage the richer list system which now supports metadata and categorization, this enable easier access via Category:Lingua Libre maintained lists > Category:Lingua Libre exclusion lists and Category:Lingua_Libre_list/Method:refreshed, therefore higher visibility and maintainability.

Remarks (4): (TBC)

Summary: A lot have been done in past 2 years despite dry funding. A lot still remains to do. The core fronts to ensure services to the community, as I suggested for over a year, are IMHO to ensure the launch of regular coding cycles to consolidate the app (multiple known issues), its ecosystem (E1 Search/Sonotheque, D4 server buffer), and implement new ideas (C1 UI change, etc.). A new team of developers have been gathered (Pushkar7077 and myself, also Aditya, SalthyKheera, Gilles), such group —the 4th one of Lingua Libre in 10 year— is fragile, there is no guarantee it will continues. A healthy dev team as well as clear planing and transparent budgeting are required to maintain the effort. Request for support should be directed to WMFR which gather grants in the name of Lingua Libre. Yug (talk) 08:24, 2 September 2026 (UTC)reply

For existing and structured wishlists, see meta (2022, updated in 2026) and phabricator. We have improvement ideas for 2 years-worth of coding. What we critically need is more developers to implement those wished features. The last coding cycle stood largely on my sabbatical volunteership. A more serious strategy and support is needed. WMFR is the recipient of two funds in the name of Lingua Libre and should lead this effort more boldly. Yug (talk) 08:04, 3 September 2026 (UTC)reply

Request to add new language to LinguaLibre

Request to add Khiamniungan language to LinguaLibre

Hello,

I am unable to add recordings for the Khiamniungan language (specifically the Patsho dialect) on this platform. Could you please enable it for us if possible?

Here are the details:

Language Name: Khiamniungan (specifically the Patsho dialect)
Wikidata ID: Q6402330
ISO 639-3 Code: kix

This will allow our community to start recording audio for this vulnerable language spoken in Nagaland, India.

Thank you for your help! ExfactoLexander (talk) 19:51, 29 July 2026 (UTC)reply

Please ask for higher userrights

@TCraze , @Нейшлотский , it appears you are very active on Lingua Libre an need more usrrights.

It's now time to go to Commons:Requests for rights page and expand your userrights.

Note: On the app server's logs we can see ...

2026-07-28 22:10:34,436 - ERROR - Failed to upload recording 35010 (file: LL-Q7737_(rus)-Irina (Нейшлотский)-Агава американская.wav): {'code': 'ratelimited', 'info': "You've exceeded your rate limit. Please wait some time and try again."
2026-07-29 18:56:24,067 - ERROR - Failed to upload recording 35387 (file: LL-Q9051_(ltz)-TCraze-amstand.wav): {'code': 'ratelimited', 'info': "You've exceeded your rate limit. Please wait some time and try again."

Yug (talk) 16:38, 1 August 2026 (UTC)reply

I'm just dropping Help:Lingua Libre/User rights#Request new user rights to help in requesting for autopatrolled. --Volvox (talk) 20:01, 1 August 2026 (UTC)reply

Hello User:Volvox, thanks for you help and sharp dive into Lingua libre processes. I actually already requested higher user rights for two users, see Commons:Requests_for_rights#TCraze. It goes faster this way. Yug (talk) 05:04, 2 August 2026 (UTC)reply

thank you! TCraze (talk) 08:50, 11 August 2026 (UTC)reply

Some bug reports

See also Help:Lingua Libre/Corrections.

After discovering Lingua Libre at this year's Wikimania in Paris, I wanted to contribute some Luxembourgish recordings with TCraze, but we encountered the following problems:

Despite having recorded 380 pronunciations, only 60 uploads got through. The other 320 recordings seem to be lost. I found the following Phabricator task, describing the same issue: T433757.
As mentioned here, even though the license was set to CC BY-SA 4.0, the structured data incorrectly indicate copyrighted, dedicated to the public domain by copyright holder (Q88088423) as copyright status (P6216). I created the task T433775 for this issue.
The files' structured data contains seemingly nonsensical values for depicts (P180). As we were using the lexeme generator, I assume this is the same issue as described in T413316.

--Volvox (talk) 20:37, 1 August 2026 (UTC)reply

Hello Volvox, TCraze,Ebene0406, Kiowap, Anup_Sadi,

Thank again Volvox for noticing this issues and reporting them.

T433757 — investigate : Yes, we found this 60 uploads limits and investigation is ongoing using Phabricator to gather information. All 5 known cases affected newly registered accounts using Lingua Libre within their first 4 days, when userrights are mostly limited. This mostlikely is some Wikimedia Commons built-in limitation, either new on their side, or which the Wikimania sprint revealed.
T433775 — license SDC data : thank for creating this task, it's the way to do so WMFR or volunteers devs take it in charge. Note: this issue is know but we lack developpers to fix it.
T413316 — other SDC data : same.

Please note this project is vastly under-funded. The 4th team of hybrid volunteer/freelance devs (myself, Pushkar, and occasionally Aditya, Gilles and others) compensated this droughts as much as possible (I personally donated a full sabbatical year). We led a massive 2024-2026 recoding, doing far more than our citizen duty and are naturally slowing down. There is still a lot to do, and more issues that those cited on this page.

WMFR has requested and will receive WMF's funding for Lingua Libre technical developements. If we do remind them, WMFR will fund Lingua Libre's feature requests such those you raise above. Yug (talk) 08:18, 4 August 2026 (UTC)reply

thank you for this! I will upload more Luxembourgish Audios soon TCraze (talk) 08:49, 11 August 2026 (UTC)reply

Hello @Aadenboy, @Volvox,

I'm focused on other matters now but here is a progress status:

(1) Limit to 60 : Still as per my previous post just above.

(2) SDC depicts: phab:T413316#12258508 is mostly solved, we now use "P973 described at url", see the link.

(3) SDC copyright: not trivial, for later.

The Django/Vuejs stack is fairly easy, if someone in your communities wants to contribute... try it.

Best regards. Yug (talk) 17:56, 26 August 2026 (UTC)reply

Thank you for the update and the partial solution. I have replaced the wrong depicts (P180) statements with appropriate described at URL (P973) statements for TCraze's uploads as suggested in the Phabricator task (giving me a good opportunity to familiarize myself with OpenRefine). --Volvox (talk) 21:18, 31 August 2026 (UTC)reply

Hello Volvox,

Relative to the non-sensical P180 on Wikidata Lexeme items, due to confusion between Qid and Lid... I hacked my way to this Sparql query which should find all records from past 6 months with a P180 `depict` and P10893 `recordist` value, then keeps the one where the ?transcription value and P180's wikidata ?depictLabel value are different. Inpecting carefully the responses I do see files, recorded via LL's Step 4 Lexemes generator, which are corrupted. I cleaned up from April to August 2026 by hand but there is a denser batch around February 5th to 11th. If there is an OpenRefine method to clean those i'm interested by it. Yug (talk) 13:23, 2 September 2026 (UTC)reply

I try to look into it this weekend. --Volvox (talk) 11:29, 4 September 2026 (UTC)reply

Missing languages from the recording language list

Hi @Yug,

Five people tried to add Emirati Arabic in order to make recordings, but we can't find it in the list of available languages. We are experiencing the same issue with Shy (Shawiya language) and Arq (Algerian Arabic).

Do you know how we can fix this or add these languages to the list? Reda Kerbouche (talk) 11:14, 13 August 2026 (UTC)reply

Removing already recorded items is not working

Hello! I've been recording some audios for the Long English words category and have added them to the English Wiktionary. I wanted to continue recording today, but I didn't want to re-record ones I'd already recorded, uploaded and used, so I tried pressing the "Remove items already recorded" button multiple times, but nothing happened. I've also turned on the toggle beforehand, and I've tried this both on my phone, with and without desktop mode, and on my laptop, but none of those options fixed anything. If I need to share anything else, please let me know. I also hope I'm adding this in the right place. UbiquitousSwizzy (talk) 18:08, 18 August 2026 (UTC)reply

2026-LinguaLibre-Step 4 to 6-remove already recorded : works, see the first 3 seconds.

Hello UbiquitousSwizzy,

My apologize for the delayed response.

The function you are referring to works on my side (see screencast on the right). This means we need more informations (dekstop/mobile, browser), or an investigation.

Wikimedia France used Lingua Libre to request funding from the WMF and the French Ministry of Culture, successfully. This should allow more coding cycles in the near future to consolidate the app as it is needed. Yug (talk) 06:46, 2 September 2026 (UTC)reply

Note: the feature is based on both the language + username + locutor name. It will not remove items recorded by another locutor. There, your locutors "Alice Rodriguez" and "Alice" and "Alice42" will be considered as 3 different locutors event if they are the same person. Yug (talk) 08:22, 3 September 2026 (UTC)reply

User:Lingua Libre Bot

See also Special:BotPasswords.

Hello everyone, User:WikiLucas00 has requested master access to User:Lingua Libre Bot. WikiLucas has been a long term contributor to Lingua Libre's on several axis : audios recording, occasional JS developer and administrator/bureacrat on our previous wiki (archive) where he also welcomed hundreds of users.

Lingua Libre Bot is critical to spread Lingua Libre's audio on ~6 wikimedia project so far. This Bot is one of the most critical bottlenecks, it currently allows and limits Lingua Libre reuse to only a few lucky communities : french, occitan, kurds, etc. Beyond WikiLucas00's handling the bot as volunteer, we are in need of deeper support to expand edition to more communities.

As for now, we can trust WikiLucas with access, exploring this tool and its code, and running when he sees fit. Yug (talk) 10:43, 2 September 2026 (UTC)reply

Datasets and TTS for Levantine Arabic

Hi all, I was thinking about building an app for language learning - think of the green owl one but open-source and better in every way. I've started with Levantine Arabic, because reasons (one of them being that it's a mostly spoken rather than written language). While trying to gather data, I came across this dataset, however I could only download a 2024 cersion from the Internet Archive.

Is a more up-to-date version of the dataset available somewhere?
Are there decent open-source Levantine Arabic TTS/STT models? Are these [2] [3] any good?
Are/were there similar projects?

Any thoughts, comments or ideas welcome. Alaexis (talk) 21:45, 6 September 2026 (UTC)reply

Availability of recent French datasets ?

Copied from Discord.

Hello / Bonjour, I'm building an offline dictionary for an e-reader based on Wiktionary entries with pronunciation audio clips, stored on device. Most of the audio I need is is yours but fetching each file from Commons will take about a week just for the french data. I am searching for per language dumps but the datasets hosted on lingualibre.org seem to be broken links.

For instance lingualibre.org/datasets/Q21-fra-French.zip just returns the site's HTML

Are archive still available somewhere? Or is there anyway to get a language in bulk ? -- anonymous, 2026.09.17

Thanks for moving this. A few findings in case they help others:

The per-language zips are still on the Wayback Machine. French (Q21-fra-French.zip, 7.81 GB) was captured 2023-09-19 and again 2025-02-24 with an identical digest, so the French build seems to have stopped around September 2023. English (980 MB) is from 2025-02-24. By June 2026 the same URLs return the site's HTML shell, so the break is somewhere in between.

For French I ended up fetching the originals from Commons directly, which took about a week under the rate limit. I'd like to do the same for English, Spanish and German, so what I'd most like to know is whether the dataset generator is still runnable, and whether there's somewhere the archives are meant to live now. Happy to help test. Redvelo (talk) 20:31, 21 September 2026 (UTC)reply

Correcting audio recordings files

See also meta:User:Yug/RenameOrReplace, Help:Lingua Libre/Corrections. To encourage community learning this discussion is shared here.

Hello User:Felino Volador,

I noticed you've been helping with Lingua Libre records.

May I encourage you to use meta:User:Yug/RenameOrReplace ? I helps me to manage mass of LL records since it allows to target a bunch of files for renaming and mass wikitext edition. In the case of Category:Lingua Libre pronunciation-quh , two things were needed :

Rename titles : moved page File:LL-Q5218 (que)-Nicaela Phuyu-ñaña.wav to File:LL-Q1307730 (quh)-Nicaela Phuyu-ñaña.wav
Edit wikitext's template value : Q5218 → Q1307730.
Edit wikitext to remove category : \n*\[\[Category:Lingua Libre pronunciation-quh\]\]\n* → « » ; (nothing=remove match).

Those 3 typical maintainances can be done on whole categories with that userscript. Yug (talk) 21:50, 6 October 2026 (UTC)reply

Hello @Yug, Thanks for the heads up! Sounds like an amazing tool. The job of correcting the audiofiles recorded in Lingualibre with the code "Q5218 (que)" has not ended, there is a lot to correct still and doing all in one go will be perfect. Is it possible to upload an image capture of a wikitext replace example? I tried but no success. How do you set up the possibility of replacing the Lingua Libre template wit the wikitext? I still have to make it manual, see diff. Cheers, Felino Volador (talk) 23:32, 6 October 2026 (UTC)reply

Hello Felino Volador,

A lot of improvements arrived this past year.

Since Pushkar and myself put online the new Lingua Libre all, metadata are immediately saved into the filepage's SCD (diff by our app, full stream, works without template parameters). So the template directly loads those and categorize the file accordingly.
With legacy, pre-May 2026 uploads, the templates and SCD were not working as well, so bots or hands edits were needed. Indeed, your diff was on a 2024's upload.

A new userscript, similar in approach to meta:User:Yug/RenameOrReplace, could automatize such edit to directly refill the SDC with the proper values.

We are too few, I'm myself on so many Lingua Libre fronts that I have difficulties to push further, but progress is on the way. Yug (talk) 08:38, 9 October 2026 (UTC)reply

Chatroom FAQ

Opening Lingua Libre project on Commons

Project structure

Announcement : New recording Studio pre-release !

fra -> sas correction needed

Reporting bugs

Bug reported by djknusper

Bug(s) report by Yug

Bug report by Lucas Werkmeister

Bug(s) report by Yug (2)

Migrating_`Translations:*`_pages

Pages to mark for translation

Audio files problem

[News] Are there broken wikipages in the project ?

Semi-protection

Help:Lingua_Libre/de is not using german

Redesign

Only_one_linked_user_allowed_on_lingualibre

Lingua Libre mqg metadata

Adding new languages to record

Q25337833 (mis) into Q7259896 (puj)

SDC fix

Bug(s) report by {Mathsou}

Unblock servers

SPARQL endpoint status

WMC API & Tatoeba.org

[Bot] Regarding Lingua Libre Bot

About custom categories

Concerning Petscan and ML voices

Swadesh lists for Indonesia languages

[Bot] Getting sound files into Wikipedia articles without doing that one by one manually?

Adding Categories

Liste de bugs ou de problèmes rencontrés

Username moving for Juwan

Audio recordings for the Sanskrit language

404 for user of a submitted audio

Archives

Bug(s) report by {Suzanne Truong WMFr}: Recordings in french

archive.lingualibre.org

New userscripts for LL maintenance

Download datasets

SDC block uses wrong copyright property

Please add Tyap dialects to Lingua Libre

Pali Q36727 and Mon-Pali Q100271134

Feedback from the Lingua Libre booth at Wikimania 2026

A) Speaker profile form

B) General UX of the Record Wizard

C) Recording

D) Uploading recordings

E) Website and codebase

Request to add new language to LinguaLibre

Please ask for higher userrights

Some bug reports

Missing languages from the recording language list

Removing already recorded items is not working

User:Lingua Libre Bot

Datasets and TTS for Levantine Arabic

Availability of recent French datasets ?

Correcting audio recordings files