Open viktara opened 5 years ago
XD, there is no end at adding a new STT or TTS. Even if kalliope already support a lot of different one, there is always a new issue to add a new one :). To answer, it is not planned, but feel free to PR the project or create a community based STT.
OK, I'll have a look into that
Is there anything other than sphinx for STT that is open source? It seems the rest are all cloud services
The only one self hosted so far.
I'm starting to work on implementing Mozilla DeepSpeech and Mozilla TTS when I get Mozilla TTS to run at all
I'm starting to work on implementing Mozilla DeepSpeech and Mozilla TTS when I get Mozilla TTS to run at all
Cool! Are you facing any issues with the deep speech side? I could put up a server for that.
@OmegaRogue Any progress ?
I would like to bring everyone's attention to this post and more importantly to this one.
@Sispheor would you consider such a change ?
diff --git a/setup.py b/setup.py
old mode 100644
new mode 100755
index 9a34e68..b957bfd
--- a/setup.py
+++ b/setup.py
@@ -75,7 +75,7 @@ setup(
'Werkzeug==0.16.1',
'pyyaml>=5.1',
'six>=1.12.0',
- 'SpeechRecognition>=3.8.1',
+ 'speech-recognition-fork>=3.8.1',
'markupsafe>=1.1.1',
'pyaudio>=0.2.11',
'pyasn1>=0.4.5',
Do we need to make a PR for it ?
Replacing a well known lib by a 2 stars fork? Just to support one more stt? I don't think that's ba good idea.
Replacing a well known lib by a 2 stars fork
The problem is that well known lib https://github.com/Uberi/speech_recognition is not maintained anymore. Pull requests are not accepted. Last commit was 2 years ago. Many important technological advances are not implemented.
Ok. Switching to a fork could be a good idea. But the owner of this particular one is saying himself that he has no experience on this. Maybe there is another fork where the major part of the community has moved already.
@Sispheor, one of the two stars is from me so I guess this is a one star fork. Looking at the "network" facility provided by github, it seems that it is the most advanced one. I'll eventually come back to this issue if I ever succeed with the kalliope and deepspeech combination.
@OmegaRogue Any progress ?
I tried for a few days to get DeepSpeech working, but didn't make any progress at all, so I gave up on it, maybe I'll try at some point
here is an example of streaming transcription from microphone in python with deepspeech
Is anyone working on or planning to support https://github.com/mozilla/DeepSpeech as a STT in kalliope?