separating tracks killed

[liabozarth]

I'm using diarize_parallel, but this following code step frequently get killed without any logs, what could be the possible issue?

        return_code = os.system(
            f'python3 -m demucs.separate -n htdemucs --two-stems=vocals "{args["audio"]}" -o "temp_outputs"'
        )

logout put:

local_file_path ./output/12325.mp3 generating transcript file ./output/12325.txt Selected model is a bag of 1 models. You will see that many progress bars per track. Separated tracks will be stored in /home/liafan/repos/local/whisper-diarization/temp_outputs/htdemucs Separating track output/12325.mp3 Killed local_file_path ./output/12327.mp3 generating transcript file ./output/12327.txt Selected model is a bag of 1 models. You will see that many progress bars per track. Separated tracks will be stored in /home/liafan/repos/local/whisper-diarization/temp_outputs/htdemucs Separating track output/12327.mp3 Killed local_file_path ./output/12329.mp3 generating transcript file ./output/12329.txt Selected model is a bag of 1 models. You will see that many progress bars per track. Separated tracks will be stored in /home/liafan/repos/local/whisper-diarization/temp_outputs/htdemucs Separating track output/12329.mp3 Killed local_file_path ./output/12331.mp3 generating transcript file ./output/12331.txt Selected model is a bag of 1 models. You will see that many progress bars per track. Separated tracks will be stored in /home/liafan/repos/local/whisper-diarization/temp_outputs/htdemucs Separating track output/12331.mp3

[MahmoudAshraf97]

probably running out of memory, execute the command on its own to debug it

[liabozarth]

you were right, the file was too large.