search

davabase / transcriber_app

Real time speech to text transcription app.
363 stars 69 forks source link

Hangs when opening stream #25

Open Kronzky opened 11 months ago

Kronzky commented 11 months ago

Thanks for putting this whole thing together! It's exactly what I needed.

Unfortunately, I can't get it to work... Neither the released exe nor the version I build myself.

Once I click on Start Transcribing I only get the "waiting circle". I'm not much of a python programmer, but it seems to get stuck at stream = pa.open(format=pyaudio.paInt16 (~line 135).

My speakers (HDMI monitor) are set up as the Stereo Mix input (which is also selected as the Audio Input), but nothing seems to arrive at the app (the volume bar is never moving). If I use my audio jack speakers as the Stereo Mix input I only get bad feedback. To disable the feedback I have to turn on "Mute the rear output device, when a front headphone is plugged in" in the Realtek Audio Console (even though, as I read elsewhere, this should be disabled if you want to use Whisper in streaming mode).

transcribe

davabase commented 11 months ago

Stereo Mix is an odd beast that I had much trouble with. One thing I did learn is that Stereo Mix will output no audio unless there is a real physical cable plugged into the speaker jack. The cable doesn't even have to be plugged into anything as far as I know.

You can try troubleshooting the Stereo Mix output by taking Transcriber out of the loop, open a audio recorder like Audacity and try recording from Stereo Mix to see if Stereo Mix is actually outputting anything.

Ultimately I ended up using a virtual cable with VB-Audio, which was also rather complicated. Based on your screenshot it looks like you may have tried that route.