RuntimeError: "slow_conv2d_cpu" not implemented for 'Half'
fantinuoli opened this issue · 5 comments
fantinuoli commented
Installed correctly on macOS Monterey. Run as indicated in Readme, after the initial prompt "cli.transcribe_from_mic INFO -> Ready to transcribe", talking into mic causes:
RuntimeError: "slow_conv2d_cpu" not implemented for 'Half'
Whisper works correctly in offline modality with CPU.
shirayu commented
Thank you for the report!
I think, this seems to be the result.
I will check this.
shirayu commented
I fixed this.
If you still have problems, please feel free to reopen this!
fantinuoli commented
Thank you for your swift reply. I see the changes having an effect, however I am still not able to run it. I do not get any reaction. Mic in Terminal is allowed.
(whisper_streaming) fc@Claudios-MacBook-Pro whisper_streaming % poetry run whisper_streaming --language en --model base -n 2 --debug
[2022-09-23 15:07:18,030] transcriber._set_dtype WARNING -> FP16 is not supported on CPU; using FP32 instead
[2022-09-23 15:07:18,030] cli.transcribe_from_mic INFO -> Ready to transcribe
[2022-09-23 15:07:18,073] cli.transcribe_from_mic DEBUG -> Segment: 0
[2022-09-23 15:07:18,498] transcriber.transcribe DEBUG -> seek=0, timestamp=0.0, rest_start=None
shirayu commented
@fantinuoli I think -n 2 is too small. How about more bigger value?