Feature: Add integrated Text To Speech based announcements
Text to Speech would be a really nice integration for announcements. Right now it needs to happen externally to the announcement. Figuring out a simple way to integrate it into announcements would make them a lot more useful.
For supporting this in conjunction with something like the coral we would need to do the following:
- [ ] Find TTS Neural net
- [ ] Test with USB attached coral (stretch goal)
- [ ] Add TTS field in announcement that is supported if attached
@Lohrer and @klay2000 thanks for the inspiration on this
Thanks really belong to @sepffuzzball in providing some possible uses for Home Assistant controlling AmpliPi https://github.com/micro-nova/AmpliPi/discussions/206#discussioncomment-2140029.
TTS anouncements would be awesome!
One potential network: ~~https://github.com/mozilla/DeepSpeech~~ https://github.com/mozilla/TTS . I wonder how much CPU it uses without a dedicated TPU?
A possible TTS implementation would be https://github.com/coqui-ai/TTS. I tried installing with pip on a Pi but it failed, looks like we'll have to do a manual install due to some of the dependencies (at least torchaudio) not providing arm binaries.