Issue with Generating f0 and hubert Files in SVC Project — Training Doesn’t Start at All on colab

[Khejel021]

Describe the bug

Hi everyone,
I’m trying to set up the SVC project for voice cloning training, but I’ve run into an issue that completely prevents the training process from starting. If anyone has faced something similar, I’d really appreciate your help.

Problem Description:
While running the pre-hubert step with the following command:

F0_METHOD = "dio" #@param ["crepe", "crepe-tiny", "parselmouth", "dio", "harvest"]
!svc pre-hubert -fm {F0_METHOD}

Even though the path to the .wav files is correct and accessible, the f0 and hubert files (like .npy) are not being generated. I’ve tested other f0 extractors too — crepe, parselmouth, harvest, and crepe-tiny — but the issue remains.

The audio files are mono and standard format.
I even trimmed the dataset down to 5 short 9-second clips for testing, but still no output. The preprocessing step just gets marked as done without actually doing anything. Then when I start training, it finishes in about 40 seconds without any error or meaningful log — the model training never actually begins.

I’ve tried setting batch size to low values like 4 or even 1, but that didn’t help either.

The .wav paths are definitely set correctly. I even tried this alternative command:

F0_METHOD = "dio" #@param ["crepe", "crepe-tiny", "parselmouth", "dio", "harvest"]
!svc pre-hubert -fm {F0_METHOD} -n 2

Still nothing.

If anyone has dealt with a similar issue or knows what might be causing this silence in both pre-hubert and training steps, I’d be grateful for your guidance.

To Reproduce

F0_METHOD="dio"
!svc pre-hubert -fm ${F0_METHOD}

This step does not generate the files necessary for training the model.

Additional context

No response

Version

4.2.26

Platform

google colab

Code of Conduct

• I agree to follow this project's Code of Conduct.

No Duplicate

• I have checked existing issues to avoid duplicates.

[truongvav113]

I am also getting the same error as you said. I have successfully trained many models before but now I don’t understand why this error appears. Have you solved this problem? If so, please help me.