[Feature request] - Integrate Voicetyping

[Ulf3000]

Given that your tool is exceptionally well-programmed and functions seamlessly across various applications, it would be beneficial to incorporate voice-typing processing through Gemini or other suitable LLMs.
It would be fantastic to have a dedicated button within your app, rather than relying on inferior voice-typing solutions. Perhaps my perspective is incorrect; if so, please correct me.

[theJayTea]

This is an interesting request that we could think about adding in the future.

There's actually a very nice dedicated model for this by OpenAI called Whisper, however, running it locally requires ~4 GB of vram/ram and almost everyone certainly wouldn't be able to run it alongside a local LLM. There's actually 1 project I found that does what you requested with this:
https://github.com/savbell/whisper-writer

A way to get free & accessible state of the art transcription would be using the Gemini API and asking the multimodal Gemini 2.0 for a transcript. However, I'm unsure what the latency would be like.

This is not something I can immediately work on, and I'd also like to hear what others think about this proposal first.