LinguaLibre:Events/2021 Wikimedia Wikimeet India

Notes are available on etherpad.wikimedia.org/p/wmwmd03s10

• Description: Presentations on Lingualibre recording contribution, and on Github opensource assets.
• Place: Zoom > https://zoom.us/j/97614383815?pwd=OW9QazdjeVFaOTF1VXQxZ0U2ZkpCUT09
• Time: 2021/02/19-21 > Sunday 21: 13h25 UTC/14h25 UTC+1.
• Organisation: Indian Wikipedia
• Presenters: User:Adélaïde Calais WMFr, user:Poslovitch, User:Yug.
  1. /Lingua_Libre_-_record_your_language
  2. /Lingua_Libre_bot_and_recording_wizard
• Participants: ~60 watchers, largely Indians leading contributors, movement leaders and developers.
• Outcome: outreach successful, to continue.
• Review & improvements: continue work on the presentation. Some proposals here.

Conference details

1. Schedule:meta:Wikimedia_Wikimeet_India_2021/Program
2. Place: Zoom.us with pre-conference trial session.
3. Checklist: meta:Wikimedia_Wikimeet_India_2021/Checklist
4. Telegram group for presenters: https://t.me/joinchat/VJ78lQaYYQTDK_Dh
5. Other forms of communication: wmwm@cis-india.org (Indian Standard Time, UTC+0530 hours).

Presentation

Presentation was mainly in 4 sections :

• Lingualibre : why & what. By Adelaide.
• Lingualibre : how to contribute / demonstration. By Adelaide.
• Lingualibre Technical side : Generalities. By Yug.
• Lingualibre Technical side : LinguaLibre:Bot roles and request process. By Poslovitch.
• Q&A. By team.

Repositories

Github.com/Lingualibre/
Repository	Technologies (Stack)	Definition and impact
Lingua-Libre-Bot	Python, autorization	Spread 400,000 audios in your wiki
RecordWizard	JS, VueJS, CSS, PHP MW-module	Mediawiki UI module to record audios/video (sign language)
LinguaRecorder	JS, NodeJS	Js library controlling audio recordings
SignIt	JS, OOJS-UI, CSS, NodeJS	Helps teach sign language
QueryViz (not presented)	SparQL	Helps extract data, files, meaning from Lili
Notes:

Post-event review

Here are few possible improvements.

General part

• Remind possible usages : Wikimedia Websites, Language preservation, Language learning, Natural language processing (text2speech, speech2text)
• The basics: this is a "recording project". We currently focus on words, but we can think and expand to more orality and knowledge related services.

Demonstration :

• (Was so complete and systematic ! Merci Adélaide !)
• by default, the limit is 380 every three days 380 per 72mins. for newly created accounts and normal accounts. See LinguaLibre:User_rights#User_rights_on_LinguaLibre for details. But! 380/hour for 4 days old accounts and an easy process to raise this significantly.
• When you change the "stop threshold" 5 to record sentences, I think (?) it's the volume level under which the audio will stop. So we have to move it down if you want longer audios.
• The "silence length" should be the single element to change.

Technical side :

• Add lingualibre.org/datasets/ & talk about external usages
• Remind possible usages : Wikimedia websites via bots, Language learning apps, Natural language processing (text2speech, speech2text).
• Referees on nearly each repository
• Welcome volunteer devs
• "We think about more features and improvements that we can develop" (therefore we need devs). See Phabricator.
• Do we have a long term strategy. What is our development vision (it's redundant with Adelaide request to define a long term )

Also:

• Calls for action. "Come and try it out".
• Visually thanks the community. (nothing in the slides!)
• Could be a final slide with the Recording Wizard screenshot.

Networking

Content Translation Tool

Lead by the WMF's Language team. See mw:Wikimedia Language engineering, meta:Category:Language research & meta:Research:Section_Translation_Design_Research.

• Hugo: Do Content Translations organizes IRL "Translathon" and collaboration with language universities ? Any feedbacks on such event ? what are the key element to kickstart such event ? -- Q by fr:Yug
  • Amir: We are a product development team: we design and development software features, and we don't organize events in our Wikimedia Foundation capacity. However, many chapters and individual community members organize such events, in many countries and for many languages. Some tips about organizing such events can be found at this link: https://meta.wikimedia.org/wiki/Best_practices_for_Content_Translation_events
  • Runa: We did a survey last year in partnership with an editathon event in India (https://outreachdashboard.wmflabs.org/courses/CT_University/Syberthon_2020_Indian_Languages/home) to understand how newer editors interacted with Content Translation. The report is here: https://meta.wikimedia.org/wiki/Research:Content_Translation_Newcomer_Survey,_India_2020 .