Support voice input

[devurandom]

It would be great if there was a button in oterm that would allow voice input, i.e. record audio, transcribe, and send the result to ollama.

There exist tools like https://github.com/DrewPear309/ollama-talk or https://github.com/maudoin/ollama-voice that use Whisper (https://github.com/openai/whisper) or Vosk (https://github.com/alphacep/vosk-api) for the transcription, but are separate tools and do not integrate into oterm. This also output the responses from Ollama, which might not always be necessary or desirable.

[ggozad]

Hey @devurandom
I actually did have audio input in the first version of oterm (not sure if it was on github at the time even).
What put me off was mostly the additional requirements and performance, so took it out.
If you can put the effort to bring it back again, I should be able to find the code I had done on that at the time. What I would like to see is

• compatibility on all platforms
• performance
• size (oterm is already bigish size-wise and don't want to make it difficult to install)
• decent ux