ASR with external API

[henk717]

Is your feature request related to a problem? Please describe.
My laptop is not as powerful as my desktop but I do want to be able to watch videos on my laptop.

Describe the solution you'd like
OpenAI (or KoboldAI API) audio transcription support, this lets me use my existing KoboldCpp instance that already runs the translations.

Describe alternatives you've considered
Remote Desktop or just waiting long / generating them on the desktop first.

Additional context
We have a free test API at https://koboldai-koboldcpp-tiefighter.hf.space/v1/audio/transcriptions which you can use for basic implementation testing (slower than self hosted).

This should also allow different speech to text engines and models if they speak the same API.

[umlx5h]

For such use, it is recommended to generate subtitles in advance with a dedicated program such as SubtitleEdit.

Audio data is large in size, which increases the amount of network traffic.
For translation, this is not a problem since it is just text data, but for audio, it is better to do it locally.

I will consider it, but very low priority.