Problem when training my own model

[Asma_Trabelsi]

Hi all

I would like to train my own model based on open source data more exactely “https://commonvoice.mozilla.org/data”.

When i launch bin/import_cv2.py --filter_alphabet path/to/some/alphabet.txt /path/to/extracted/language/archive, I have faced the problem
FileNotFoundError: [Errno 2] No such file or directory: ‘fr/clips/train-all.csv’

any solution please ?

[othiele (Olaf Thiele)]

Please read the script to understand more, this is still work in progress. And as usual, read the output. It looks in the “fr” directory, which should work for French data. But as you didn’t supply us with more info, I don’t know whether this should be fine:

[READ FIRST] What and how to report if you need support

We are happy for every new user of DeepSpeech. As it is under active development, stuff changes quickly and you might have problems running it. We are here to help and might point to this thread if needed Please do the following before posting: Search this forum or Google for your problem Read the documentation for your version here Check the new Playbook on how to use DeepSpeech Check Github issues for more technical problems If you still struggle, feel free to open a new t…

[Asma_Trabelsi]

Thank you for your reply. My problem is that the downloaded data does not contain the file train-all.csv in the clips folder and there is no idea how to create this file (it is not mentionned in the doc

[othiele (Olaf Thiele)]

Try audiomate, it takes several inputs and can produce valid DeepSpeech data.

Or read the import script, either way it is a bit of work. The train is just a csv with all the files for training, but you also need a dev.

[lissyx ((slow to reply) [NOT PROVIDING SUPPORT])]

@Asma_Trabelsi

bin/import_cv2.py --filter_alphabet path/to/some/alphabet.txt /path/to/extracted/language/archive

Is that the exact command line you used?

[Asma_Trabelsi]

Exactely i have used bin/import_cv2.py /path/to/extracted/language/archive
when i have replaced the path by the french downloaded model.

i have no idea about the alphabet.txt. it does not exists in the downloaded model

[lissyx ((slow to reply) [NOT PROVIDING SUPPORT])]

You obviously have not read the documentation. You need to generate it.
Please, it really looks like you should rely on https://github.com/Common-Voice/commonvoice-fr/blob/master/DeepSpeech/Dockerfile.train if you really want to train a french model, or directly use pre-trained ones.