Can't start training due to recursion depth error

[yukiarimo]

Describe the bug

I have manually split my .wav audio file into chunks of 10 seconds in 75 files and written transcriptions for them. However, when I tried to train the model I got an error: RecursionError: maximum recursion depth exceeded while calling a Python object. Nevertheless, I have used a similar approach to another dataset with 100 audio chunks with the same duration, and it worked.

Can you please help to fix it? Is it possible to use that amount of data? I also tried to duplicate all samples to make 150. However, it didn't helped!

A similar issue has been seen and not resolved: #3410

To Reproduce

Try to train on the songs

Expected behavior

Should work

Logs

No response

Environment

SAME ON THE INSTRUCTION

Additional context

No response