-
-
Notifications
You must be signed in to change notification settings - Fork 1.3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
feat: simple coqui-ai stt integration #341
Conversation
scripts/check.js
Outdated
} else { | ||
log.success(`Found DeepSpeech language model at ${deepSpeechPath}`) | ||
log.success(`Found DeepSpeech language model at ${coquiPath}`) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
nit: s/DeepSpeech language model/Coqui acoustic model/
Coqui (like DeepSpeech) has both an acoustic model and a language model. The acoustic model is model.tflite
and the language model ends in *.scorer
. Here you're checking for coquiPath
, which has been set to the acoustic model
scripts/setup-offline/setup-stt.js
Outdated
const tmpDir = 'scripts/tmp' | ||
const deepSpeechVersion = '0.9.3' | ||
const coquiVersion = '1.0.0' |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
s/1.0.0/1.1.0/
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
i couldn't find a v1.1.0
here: https://github.com/coqui-ai/STT-models/tree/main/english/coqui, so i've renamed the variable to coquiModelVersion
to distinguish it from the release version.
server/src/stt/coqui-stt/parser.js
Outdated
|
||
/* istanbul ignore next */ | ||
try { | ||
STT = require('stt-gpu') // eslint-disable-line global-require, import/no-unresolved |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Coqui only releases TFLite models, so there's no GPU needed. stt-gpu
is depreciated. You can read more detail here.
@@ -10,7 +10,8 @@ class Stt { | |||
this.providers = [ | |||
'deepspeech', |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
s/deepspeech/coqui/
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM! Thanks for integrating Coqui STT! :)
Hi @johannbarbie 👋, I just tried and it works out of the box. That is neat, thanks 👍 The docs and new landing page will be updated in the coming minutes according to these changes. |
New landing page looks slick 😊👍 Is there somewhere I should upload a high res logo for Coqui? |
Thanks! Would it be possible to send a SVG logo of Coqui on Discord? |
All nodes have been updates as mentioned in #329 (comment). @JRMeyer I used this SVG and added the Coqui color to it. Please let me know if you still prefer to change it. Same as the docs, if you'd like to add/change something, do not hesitate 😉 |
What type of change does this PR introduce?
Does this PR introduce breaking changes?
List any relevant issue numbers:
Description:
This adds Coqui for offline STT. The parser is a copy of the deepspeech version. It works for me with node 16 and mac.
I have no prior experience with Leon or Coqui STT, so this might be very irrelevant or incomplete, let me know.