Question regarding the future of this project

[est7]

Hi @slopus and the team,
First of all, I want to express my sincere gratitude for this project. Happy is an amazing tool/library, and it has been incredibly helpful to me. I truly love the work you've put into it.
I've noticed that the repository has been relatively quiet recently, with the last commit being about two months ago. I also see several PRs and Issues pending review.
I completely understand that open-source maintenance takes a lot of time and effort. I wanted to respectfully ask about the current status of the project:

1. Is the project still being actively maintained?
2. Are there any plans for a commercial shift or a new business model that we should be aware of?
   It would be a significant loss to the community if this project were to stall, given how excellent it is. Knowing the roadmap would help me (and others) decide how to proceed with our integration.
   Thank you again for your hard work!

[GrocerPublishAgent]

Happy is back 🤠.

I've started merging PRs and getting all the issues connected up. @leeroybrun has been added as a maintainer. He has implemented probably 100 issues worth of features in his leeroy-wip branch. The intention is to merge and release this week an over the air update. Probably thursday?

No business model other than changing the voice agent feature to require putting in your own Eleven Labs or other provider API Keys.

I hear that @bra1nDump might have lost an entire vacation's worth of money on accident by not paying attention to the auto billing over the holidays. So we all owe him thanks for this Christmas present, but the free voice mode won't continue.

If you have played with local STT and TTS models then they are pretty good. So hopefully we can integrate those in the happy-server so that everyone does not have to pay cloud providers. I lack the technical expertise on this subject to be able to give concrete guidance on what needs to be implemented to make self hosted voice happen.

[maxfi]

Thanks a ton for the update and for all the work getting Happy back into shape 🙏❤️
I use Happy daily and it’s made me noticeably more productive. Big thanks as well to @bra1nDump for funding/letting us test the voice integration. It’s been an amazing feature to try. 🥳
Totally understand the move to “bring your own API key” for voice. Looking forward to the release this week!

[rudolfjs]

I think being your own key for voice makes complete sense. I'm a little worried about Anthropric banning Claude Code wrappers but that's a different issue.

[ProgenyAlpha]

What voice and personality model do you use? I really like the happy voice and personality; it's not too verbose, and it's really good at giving Claude what it needs to do well.

[SuperPauly]

On device TTS & STT pretty good now. With TTS models like:

• Supertonic 2
• Kokoro
• MioTTS-0.1B
• NeuTTS Nano
  Realtime CPU inference is possible on phones.

As for STT the models are a bit larger around 150mb – 300mb but mostly have ONNX quants:
– distil-whisper/distil-small.en

• UsefulSensors/moonshine-streaming-tiny
• nvidia/canary-180m-flash
• nvidia/parakeet-tdt-0.6b-v2
• nvidia/parakeet-tdt-0.6b-v3

Parakeet has ONNX quant's and a great RTFx & low WER.

Though not sure if this translates into CPU. I wonder if you can use WebView and then call the WebGPU to transcribe for GPU acceleration. Basically this https://github.com/xenova/whisper-web?tab=readme-ov-file but via WebView.

Or go the quickest route and use https://github.com/ggml-org/whisper.cpp/tree/master which has Android and iOS support.