synesthesiam / voice2json

Command-line tools for speech and intent recognition on Linux

MIT License

1.09k stars 63 forks source link

Roadmap #44

Open synesthesiam opened 3 years ago

synesthesiam commented 3 years ago

Development Roadmap for voice2json

v2.1

[X] Upgrade to DeepSpeech v0.9
[X] Automatic language file downloads
[X] New Kaldi/DeepSpeech profiles (Czech, German, French, Italian, Russian, Swedish)

v2.2

[ ] Swap Mycroft Precise for Raven (wake word)
[ ] Swap MaryTTS for Larynx (text to speech)

v2.3

[ ] Dynamic grammars for Kaldi models (change active intents for ASR + NLU w/o re-training)
[ ] Generate slot graphs separately as Kaldi grammar FSTs
[ ] Use word-level confidence values from Kaldi to estimate sentence confidence
[ ] Investigate acoustic distance between input audio and output sentence as a confidence measure

atakangah commented 9 months ago

I was able to build from source on MacOS 14 Sonoma. I can still put up PR for the changes I made to make it work, right?