Whisper Windows CLI not producing any output?

[YutMarma]

After installing Whisper, when I run it on a test file I get no output at all.

whisper testaudio.wav

Hangs for a few seconds, then gives me a fresh prompt. No files are created anywhere. I've also tried with a testvideo.mp4 - same result: a brief pause, then back at C:> (or, in this case, U:>, since that's where I'm running whisper from, but, you know...)

Explicitly requesting a format (--output_format txt) doesn't change this behavior either.

The first time I ran whisper it did download the models, and I've tried running it in an admin command prompt in case there were weird permissions issues.

Is there a local log I can find in order to see exactly what's going on?

[glangford]

Is there a local log I can find in order to see exactly what's going on?

At a minimum, running in the console with no flags we should expect to see
UserWarning: FP16 is not supported on CPU; using FP32 instead

so something fundamental is wrong.

Some things to try:

• Run whisper --help to verify you are really running this whisper and not something else.
• Confirm you have ffmpeg installed correctly
• Try running whisper with a file name that does not exist and ensure that it complains

Running whisper with a non-existent filename should produce something like this:

  warnings.warn("FP16 is not supported on CPU; using FP32 instead")
Traceback (most recent call last):
  File "/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/site-packages/whisper/audio.py", line 58, in load_audio
    out = run(cmd, capture_output=True, check=True).stdout
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/subprocess.py", line 571, in run
    raise CalledProcessError(retcode, process.args,
subprocess.CalledProcessError: Command '['ffmpeg', '-nostdin', '-threads', '0', '-i', 'xxx.moo', '-f', 's16le', '-ac', '1', '-acodec', 'pcm_s16le', '-ar', '16000', '-']' returned non-zero exit status 1.

[YutMarma]

Hi Glenn, thanks for taking the time to look into this for me.

Run whisper --help to verify you are really running this whisper and not something else.

U:\>whisper --help
usage: whisper [-h] [--model MODEL] [--model_dir MODEL_DIR] [--device DEVICE] [--output_dir OUTPUT_DIR]
               [--output_format {txt,vtt,srt,tsv,json,all}] [--verbose VERBOSE] [--task {transcribe,translate}]

(etc etc)

Confirm you have ffmpeg installed correctly

U:\>ffmpeg
ffmpeg version 6.0-full_build-www.gyan.dev Copyright (c) 2000-2023 the FFmpeg developers
  built with gcc 12.2.0 (Rev10, Built by MSYS2 project)

(etc etc)

Try running whisper with a file name that does not exist and ensure that it complains

U:\>dir nosuchfile.wav
 Volume in drive U is USEEME
 Volume Serial Number is E851-F1CF

 Directory of U:\

File Not Found

U:\>whisper nosuchfile.wav

c:\Users\YutYut\AppData\Roaming\Python\Python310\site-packages\whisper\transcribe.py:126: UserWarning: FP16 is not supported on CPU; using FP32 instead
  warnings.warn("FP16 is not supported on CPU; using FP32 instead")
Traceback (most recent call last):
  File "c:\Users\YutYut\AppData\Roaming\Python\Python310\site-packages\whisper\audio.py", line 58, in load_audio
    out = run(cmd, capture_output=True, check=True).stdout
  File "c:\Users\YutYut\AppData\Local\Programs\Python\Python310\lib\subprocess.py", line 526, in run
    raise CalledProcessError(retcode, process.args,
subprocess.CalledProcessError: Command '['ffmpeg', '-nostdin', '-threads', '0', '-i', 'nosuchfile.wav', '-f', 's16le', '-ac', '1', '-acodec', 'pcm_s16le', '-ar', '16000', '-']' returned non-zero exit status 1.

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "c:\Users\YutYut\AppData\Roaming\Python\Python310\site-packages\whisper\transcribe.py", line 597, in cli
    result = transcribe(model, audio_path, temperature=temperature, **args)
  File "c:\Users\YutYut\AppData\Roaming\Python\Python310\site-packages\whisper\transcribe.py", line 133, in transcribe
    mel = log_mel_spectrogram(audio, model.dims.n_mels, padding=N_SAMPLES)
  File "c:\Users\YutYut\AppData\Roaming\Python\Python310\site-packages\whisper\audio.py", line 140, in log_mel_spectrogram
    audio = load_audio(audio)
  File "c:\Users\YutYut\AppData\Roaming\Python\Python310\site-packages\whisper\audio.py", line 60, in load_audio
    raise RuntimeError(f"Failed to load audio: {e.stderr.decode()}") from e
RuntimeError: Failed to load audio: ffmpeg version 6.0-full_build-www.gyan.dev Copyright (c) 2000-2023 the FFmpeg developers
  built with gcc 12.2.0 (Rev10, Built by MSYS2 project)

(snipping spam after that)

I suspect that this might be due to the early 13th-gen Intel processor I'm on*, but would like some way of confirming that the process is silently failing as a result of random segfaults rather than user-error. Is there a verbose logging option somewhere I can enable?

*(tl;dr is 13th/14th gen Intels were sort of suicidal on release, a microcode patch was rolled out that would prevent them from doing this but could not repair any existing physical damage that already occured, and this results in some rather nonsensical/unpredictable instability under heavy CPU load - I've had issues with tesseract which I'm 99.9% sure were the result of this)

[glangford]

Is there a verbose logging option somewhere I can enable?

For whisper, --verbose True is the default. I'm not familiar with any past examples of whisper silently failing like you have described.

If you suspect the processor, have you run a diagnostic on it, eg -

The Intel® Processor Diagnostic Tool
https://www.intel.com/content/www/us/en/support/articles/000005567/processors.html#:~:text=The%20Intel®%20Processor%20Diagnostic%20Tool%20or%20Intel®%20PDT,Verify%20the%20processor%20operating%20frequency.

https://www.intel.com/content/www/us/en/download/15951/intel-processor-diagnostic-tool.html

Perhaps this link is also helpful
https://www.reddit.com/r/overclocking/comments/1axepvu/optimizing_stability_for_intel_13900k_and_14900k/