Bug: Failed to read audio file

[misters2008]

Contact Details

[email protected]

What happened?

Hi, i ve downloaded whisper-large-v3.llamafile and added ".exe" to it. It opened a command window and showed me "whisper server listening at http://127.0.0.1:8080" i opened tis link in browser and saw an interface to upload an audio file and set a temperature.

i uploaded 10mb .mp3 audiofile and when i submitted it, it showed tha page with only a single phrase: "Failed to read .Wav file"

Meanwhile in the opened command window it is showing this: "Received request: Howcut.mp3 ID3: converting to wav... ID3: failed to open audio file: Invalid argument (we support .wav, .mp3, .flac, and .ogg) error: failed to read WAV file"

how to tackle this?

Version

dont know how to check, since i donwloaded a single .llamafile and dodnt download a separate app. the opened llamafile.exe doesnt let me enter anything into the command line.

What operating system are you seeing the problem on?

Windows

Relevant log output

No response

[WilliamTambellini]

same issue here on linux.

[WilliamTambellini]

whisperfile-0.8.13 seems to work though:

 $ ./whisperfile-0.8.13 -m ggml-large-v3.bin -f jfk.wav 
whisper_init_from_file_with_params_no_state: loading model from 'ggml-large-v3.bin'
...
[00:00:00.300 --> 00:00:09.000]   And so, my fellow Americans, ask not what your country can do for you, ask what you
[00:00:09.000 --> 00:00:11.000]   can do for your country.

[gnmarten]

AFAIK, wav file needs to be in a specific format: convert via ffmpeg -i "your mp3 or whatever file" -ar 16000 -ac 2 -c:a pcm_s16le -y output.wav (I did not have success with the --convert flag within whisperfile itself)

[cjpais]

this should be resolved with the latest code of whisperfile in #568 and the following commit 74dfd2

[montvid]

I got some audio files from a dvd vob file extracted with ffmpeg and got this error. Tried to extract the vob file to mkv and got he same error. Now openwhisper and fasterwhisper use ffmpeg and work fine but your implementation uses something else... other files work fine just the audio extracted from vob do not. ffmpeg -fflags +genpts -i combined.vob -c:v copy -c:a copy output.mkv ffmpeg -v error -i ~/Videos/output.mkv -f null -

[null @ 0x562429b9a280] Application provided invalid, non monotonically increasing dts to muxer in stream 1: 252033024 >= 252033024

./whisperfile-0.8.16 -m whisper-large-v2.bin --gpu nvidia -pc -f test.WAV

test.WAV: failed to read pcm frames from audio file: At end otalerror: failed to read audio file 'test.WAV'

[pawel665j]

Is this for me? Maybe I'm already too carried away by AI. Anyway, thanks for the connection and support.

---

From: montvid @.> Sent: Saturday, November 30, 2024 12:33 AM To: Mozilla-Ocho/llamafile @.> Cc: Subscribed @.***> Subject: Re: [Mozilla-Ocho/llamafile] Bug: Failed to read audio file (Issue #556)

I got some audio files from a dvd vob file extracted with ffmpeg and got this error. Tried to extract the vob file to mkv and got he same error. Now openwhisper and fasterwhisper use ffmpeg and work fine but your implementation uses something else... ffmpeg -fflags +genpts -i combined.vob -c:v copy -c:a copy output.mkv

test.WAV: failed to read pcm frames from audio file: At end otalerror: failed to read audio file 'test.WAV'

— Reply to this email directly, view it on GitHubhttps://github.com/Mozilla-Ocho/llamafile/issues/556#issuecomment-2508680415, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AOOFRCFIR7BVQLOPQM57C6L2DDMS7AVCNFSM6AAAAABNGPGW6OVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKMBYGY4DANBRGU. You are receiving this because you are subscribed to this thread.Message ID: @.***>

[cjpais]

Without seeing your WAV file directly, it's hard to replicate this issue.