Mobile-Artificial-Intelligence / maid

Maid is a cross-platform Flutter app for interfacing with GGUF / llama.cpp models locally, and with Ollama and OpenAI models remotely.
MIT License
1.51k stars 165 forks source link

TTS chat feature request #244

Closed x-legion closed 4 months ago

x-legion commented 10 months ago

Implement Android native TTS for dyslexic people. Kind of like how Google Bard is doing.

danemadsen commented 10 months ago

not sure if Android TTS works on all platforms and im not really interested in implementing too much platform specific code. But ill look into it

Ar57m commented 10 months ago

would be nice, I would love to use it with sherpa-onnx

danemadsen commented 10 months ago

I can use https://pub.dev/packages/flutter_tts and it will be ok for all platforms (excluding linux maybe) but to be perfectly honest im kind of waiting for there so be a decent open source alternative to something like eleven labs that i can implement

x-legion commented 10 months ago

I can use https://pub.dev/packages/flutter_tts and it will be ok for all platforms (excluding linux maybe) but to be perfectly honest im kind of waiting for there so be a decent open source alternative to something like eleven labs that i can implement

I use this app primarily on Android only. Windows and Linux have better alternatives to this. Maybe you can fork some code from here for Android. https://github.com/jing332/tts-server-android

baronvertigovongrahamthesecondofsealand commented 7 months ago

This would be a cool feature. Being able to speak in the keyboard (in the android app) is great, but having the app speak back would be killer.

tangyiyong commented 7 months ago

openai's speech api, or microsoft's edge-tts, better than flutter_tts.

lee-b commented 4 months ago

Worth noting that LocalAI supports basically all of the OpenAI API, but locally. That includes multiple models, text to speech, speech to text, image generation, and image recognition. It would make sense to use that, as a starting point. The combination of running a very small model locally on mobile + LocalAI for a more powerful model when remote PC connectivity is available would be great.

davidmigloz commented 4 months ago

sherpa-onnx has now an official Dart client:

danemadsen commented 4 months ago

@davidmigloz Sounds great all integrate that for the 1.3 release

danemadsen commented 4 months ago

doesnt seem to have out of the box support for flutter TTS yet? only in dart and it needs me to provide my own piper binaries

davidmigloz commented 4 months ago

I haven't tried it myself yet, but yes, it's not "out of the box", you have to manage the TTS models yourself.

danemadsen commented 4 months ago

@davidmigloz

Managing the models isnt a problem, its managing the binaries. It doesnt appear they have ffigen set up correctly to compile the required C++ code.

This would mean i would need to obtain the correct precompiled binaries for Android x86_64 and arm64, Linux, Windows and now MacOS x86_64 and arm64.

Then i would need to ship these precompiled binaries with maid which is something im not willing to do because its against open source and it would cause issues with f-droid.

I am still slowly working on the TTS question but yeah i need it to compile on build

davidmigloz commented 4 months ago

ahh indeed, I see what you mean

danemadsen commented 4 months ago

babylon_tts is working now