Not able to generate train.txt file to run multi-speaker libritts for FastSpeach 2

[hasnin24]

I tired to use MFA based extraction for FastSpeech2 on Windows 10 but I kept getting:

OSError: [WinError 193] %1 is not a valid Win32 application 

when running:
  python examples/mfa_extraction/run_mfa.py \
    --corpus_directory ./libritts \
    --output_directory ./mfa/parsed \
    --jobs 8

I was wondering what is the format for train.txt. In the comments section for /examples/fastspeech2_libritts says the format should be like this:

Wav Path | Text | Speaker Name
Wav Path2 | Text | Speaker Name

however, no matter what I do I keep getting an error when I preprocess my data using this command:

 tensorflow-tts-preprocess --rootdir ./libritts --outdir ./dump_libritts --config preprocess/libritts_preprocess.yaml --dataset 
 libritts --n_cpus 16

The error when I run the above command:

  File "c:\users\user\anaconda3\envs\tts\lib\multiprocessing\pool.py", line 867, in next
      raise value
  KeyError: '@he'

sample of the train.txt I created following the recommended format:

55/LJ001-0002 | in being comparatively modern | 55
22/LJ048-0006 | or to the vice president, end quote | 22

[ZDisket]

@hasnin24 I see you're using Windows, you should try Linux.

[hasnin24]

@hasnin24 I see you're using Windows, you should try Linux.

I was able to run it on google colab!!! Thanks for replying. I do have one quick question, is there any way I add a new voice to an existing one for FB2 if so is there any docs to follow? thanks again.

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.