Vivek Uppal
Vivek Uppal
User should be able to choose different content strings on the left side and any corresponding responses from LLM should be displayed on the right side again.
### Discussed in https://github.com/vivekuppal/transcribe/discussions/192 Originally posted by **tan-banana** April 4, 2024 similar to how open ai's backend has the option to edit context and other model variables. Say I get...
Currently the response language of LLM can be set in the parameters.yaml. Allow user to set the language using UI. Users choice should be persisted across restarts of the application.
Transcribe currently only works on Windows. Enable transcribe to work on Mac as well.
The application is way more useful if users are able to create and customize their own prompts. Provide a way in the UI to customize prompts and persist the original...
If API key is not provided, some of the buttons that depend on backend APIs are disabled in the UI. Add a tooltip to buttons to illustrate that buttons depend...
whisper.cpp based transcription currently loads the model in memory for every iteration of the transcription. This can be easily improved by loading the model in memory only once. Saves a...
Faster-whisper is touted to be the fastest offline STT derivative of openai whisper. https://github.com/SYSTRAN/faster-whisper Incorporate that into transcribe and figure if that is the best offline STT available to us.
Quite a few things to be accomplished in this PR