You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I built whisper.cpp with -DWHISPER_FFMPEG=ON and tried to transcribe an MP4 file with a command like .../whisper.cpp-1.7.3/bin/main -m .../ggml-large-v3-turbo-q8_0.bin -of Speech -otxt -osrt Speech.mp4 -l auto
I got a lot of errors, and only the first 7 seconds out of 2 minutes were transcribed. The typical FFMPEG errors were:
[aac @ 0x55cd235cfb00] TNS filter order 29 is greater than maximum 12.
[aac @ 0x55cd235cfb00] Number of bands (26) exceeds limit (2).
[aac @ 0x55cd235cfb00] skip_data_stream_element: Input buffer exhausted before END element found
[aac @ 0x55cd235cfb00] Prediction is not allowed in AAC-LC.
[aac @ 0x55cd235cfb00] Scalefactor (-1) out of range.
[aac @ 0x55cd235cfb00] invalid band type
[aac @ 0x55cd235cfb00] Gain control is not implemented. Update your FFmpeg version to the newest one from Git. If the problem still occurs, it means that your file has a feature which has not been implemented.
[aac @ 0x55cd235cfb00] channel element 2.5 is not allocated
[aac @ 0x55cd235cfb00] Sample rate index in program config element does not match the sample rate index configured by the container.
[aac @ 0x55cd235cfb00] decode_pce: Input buffer exhausted before END element found
[aac @ 0x55cd235cfb00] Reserved bit set.
[aac @ 0x55cd235cfb00] channel element 2.2 is not allocated
[aac @ 0x55cd235cfb00] Pulse data corrupt or invalid.
[aac @ 0x55cd235cfb00] skip_data_stream_element: Input buffer exhausted before END element found
The same MP4 was correctly converted to WAV without errors by the command ffmpeg -i Speech.mp4 -ar 16k out.wav
The full console output of whisper.cpp is attached as log1.txt.
Also note that out of 6 transcribed lines only the first 3 are correct. The text beyond 00:00:06.420 is some garbage that doesn't appear anywhere in the transcribed file.
The text was updated successfully, but these errors were encountered:
I built whisper.cpp with
-DWHISPER_FFMPEG=ON
and tried to transcribe an MP4 file with a command like.../whisper.cpp-1.7.3/bin/main -m .../ggml-large-v3-turbo-q8_0.bin -of Speech -otxt -osrt Speech.mp4 -l auto
I got a lot of errors, and only the first 7 seconds out of 2 minutes were transcribed. The typical FFMPEG errors were:
The same MP4 was correctly converted to WAV without errors by the command
ffmpeg -i Speech.mp4 -ar 16k out.wav
The video was downloaded from https://t.me/botcharov/11840.
OS: Gentoo Linux, AMD64.
whisper.cpp version 1.7.3.
The full console output of whisper.cpp is attached as log1.txt.
Also note that out of 6 transcribed lines only the first 3 are correct. The text beyond 00:00:06.420 is some garbage that doesn't appear anywhere in the transcribed file.
The text was updated successfully, but these errors were encountered: