Open juanfradb opened 1 year ago
That would be nice. Should be faster to speak out your prompts than to write them out.
similar to #616
There's WhisperWriter for input. It has the advantage of being able to run locally, without the need to send your every word over the internet.
On Windows 10, the TTS voices are decent, and you could make great use of the TTS engine. Ideally, there would be skip-forward and skip-backward buttons that would skip sentence-wise (and start the utterance anew). Code blocks should be ignored. The utterance could be started at the selection or the start of the sentence containing the selection. In browsers, the selection can be zero-length by simply clicking somewhere once - you'd click on the text and start play, when the starting point shouldn't be the beginning of the answer just generated.
nearly a year in a zero progress on this? honestly man if this project was open sourced I could implement that within a few hours...
I think adding support for a whisper.cpp or other local version would be helpful especially with taking notes
It would be great if one could chat directly with voice and then get back in voice with the new release of TTS in the Devday conf