Voice training on a Raspberry PI on piper-recording-studio issue

[ciaotesla]

I follow this instructions: https://ssamjh.nz/create-custom-piper-tts-voice/ to run piper-recording-studio (similar to https://github.com/rhasspy/piper-recording-studio)

When I execute the command:
python3 -m export_dataset output/en-US/ my-dataset/

I get the following warning and in my-dataset:
WARNING:main:Missing text file: output/en-US/8c576e13-b9d3-4a21-8fd6-25de426e473b.txt WARNING:main:Missing text file: output/en-US/e7b39846-a951-4d78-9b09-2f81473c4235.txt WARNING:main:Missing text file: output/en-US/5a1f0bec-3692-4824-9ab3-f9d3df446b4f.txt WARNING:main:Missing text file: output/en-US/a517017c-c54b-49ca-bf49-faaf0b6b139b.txt WARNING:main:Missing text file: output/en-US/b4427aab-17e8-4215-b867-4290c9734ef8.txt

In the my-dataset folder there is a empty metadata.csv.

In my output/en-US/ I have the recordings from the recording studio web interface.

Am I missing something? where I get those txt files? I am assuming they are generated when I record my voice?
thx

[ciaotesla]

I got the command executing correctly.
Now how I convert mydataset to a model .onxx and .onxx.jason that I can use with Piper ?
Is there a too?
thx