I’m working on setting up the back end and there’s an entry for Google API for STT. I have deep speech working that is hosted on localhost. Could I somehow use deepspeech or a different STT with the backend?
Start here. Disclaimer: Lots of people have done this, but I haven’t, I’m just driving by with a link drop to docs. That said, there are chat forum residents who know it very well. I mention all this because anything still labeled Mozilla DeepSpeech is at least a few weeks out of date, but there are definitely people using Coqui right now, so if the instructions don’t work, speak up =P
Yeah the DeepSpeech config should work - definitely let us know if it doesn’t.
For anyone finding this trying to get Coqui STT support - my understanding is that the Coqui STT doesn’t use the same interface as the DeepSpeech STT. So whilst you can use Mozilla TTS config for Coqui TTS - you can’t use DeepSpeech for Coqui STT. It’s going to need it’s own plugin.
The deepspeech settings in the core package work. I am wondering about the selene-backend calling to the google STT API. There is this deep speech server that can be made locally with a call to an API GitHub - buddyeorl/deep-talk: Deep-speech react app to test trained models,to visualize the speech to text process, to record the audio from mic to wav using the webaudio API, or to create/use a custom open speech-to-text API.
Could I add API keys to this app and use it in place of the Google STT API call in the selene-backend settings?
Hi @spatial25 if you are using your own instance of Selene then yes you could add an endpoint to point to your DeepSpeech instance but it’s not something that is built into Selene at the moment.
If it’s just for your own purposes, you’re probably better off pointing your individual Mycroft devices to DeepSpeech.