Hi all,
In my previous report it was mentioned that my project for second phase of
GSoC was revised. I feel like I have made decent progress the previous two weeks towards implementing a speech-to-text service in
Jitsi-Meet.
I spend the start of week 6 on getting familiar with _javascript_ and the
jitsi-meet and lib-
jitsi-meet projects, as well as find out how to record the audio from a conference in
jitsi-meet. At the end of the week I managed to write some simple recording functionality[1].
When the audio streams were able to be stored. the next step was to determine the best way of transcribing the audio. We determined that the best option was to use the Sphinx4 speech-to-text library. This week I created a HTTP server[2] to which jitsi-meet will send audio fragments. The server will than use the Sphinx4 library to get the text in the audio fragments and send it back.
Regards,
Nik
[1] = https://github.com/nikvaessen/jitsi-meet/blob/audioRecording/modules/audioRecording/audioRecorder.js
[2] = https://github.com/nikvaessen/Sphinx4-HTTP-server