search

nextcloud / spreed

🗨️ Nextcloud Talk – chat, video & audio calls for Nextcloud
https://nextcloud.com/talk
GNU Affero General Public License v3.0
1.64k stars 437 forks source link

Generate automatic subtitles #3969

Closed jospoortvliet closed 4 years ago

jospoortvliet commented 4 years ago

Summary

Sometimes, you're in a call with a bunch of Germans. Or French people. And you might not be German. Or French. At that point, while you could all speak Esperanto or English. But that isn't always possible! For such a situation, the automatic creation of subtitles would be immensely helpful.

This might not be super easy, on one's own server, but then again, who knows, it is 2020 already ;-)

Solution

User can enable subtitles and select a language from a drop-down or such. Then, subtitles appear on the bottom of the video of the speaker.

Alternative

Learn every language in the world

nickvergessen commented 4 years ago

Feel free to reopen when there is a JS based voice transcription lib available

juliusknorr commented 1 year ago

Not sure how feasible but whisper.cpp seems to be able to be compiled to WebAssembly: https://github.com/ggerganov/whisper.cpp/tree/master/examples/whisper.wasm

sunjam commented 1 year ago

Feel free to reopen when there is a JS based voice transcription lib available

Would Google Live Transcription be an option? It is coded in Java and C and available under Apache 2.0 license.

nickvergessen commented 1 year ago

Well that basically means we would need yet another service that admins run. The thing Julius found would run on the user's browser? Which would be much better

simonspa commented 1 year ago

Would Google Live Transcription be an option? It is coded in Java and C and available under Apache 2.0 license.

The repository says

This repository contains the Android client libraries for communicating with Google's Cloud Speech API that are used in Live Transcribe

so as far as I understand the actual transcription happens in a closed-source online service by Google and this is just the Android library to make use of their API.