Whisper Command never processes audio

[TheOriginalBytePlayer]

I got it built and it runs but when I say the keyword phrase it calls transcribe and never ever returns from it.

Any ideas what's going on?

[danbev]

I've tried this on Linux (ubuntu 24.04) and I'm able to say the keyword phase:

main: processing, 4 threads, lang = en, task = transcribe, timestamps = 0 ...

init: found 1 capture devices:
init:    - Capture device #0: 'EtronTech CMOS based eSP570 WebCam [Onyx Titanium TC101] Mono'
init: attempt to open default capture device ...
init: obtained spec for input device (SDL Id = 2):
init:     - sample rate:       16000
init:     - format:            33056 (required: 33056)
init:     - channels:          1 (required: 1)
init:     - samples per frame: 1024

process_general_transcription: general-purpose mode

process_general_transcription: Say the following phrase: 'Ok Whisper, start listening for commands.'

process_general_transcription: Speech detected! Processing ...
process_general_transcription: Heard 'Okay Whisper, stop listening for commands.', (t = 674 ms, p = 58.61%)

process_general_transcription: The prompt has been recognized!
process_general_transcription: Waiting for voice commands ...

process_general_transcription: Speech detected! Processing ...
process_general_transcription:   DEBUG: txt = 'Okay Whisper, stop listening for commands. Turn left.', prob = 49.53%
process_general_transcription: Command 'Turn left.', (t = 769 ms)

process_general_transcription: Speech detected! Processing ...
process_general_transcription:   DEBUG: txt = 'Okay Whisper, stop listening for commands. Turn right.', prob = 44.17%
process_general_transcription: Command 'Turn right.', (t = 733 ms)

process_general_transcription: Speech detected! Processing ...
process_general_transcription:   DEBUG: txt = 'Okay Whisper, stop listening for commands. Move forward.', prob = 33.53%
process_general_transcription: Command 'Move forward.', (t = 743 ms)

But I do have issues sometimes with my wireless headset and that the microphone does not always work. Can you show us the output you are seeing?

[TheOriginalBytePlayer]

Okay, the issue appears to have been the size of the model and the optimization of the command.app Once I rebuilt it in release mode, added vulkan to the build, and quantized the tiny model, it became much more responsive.