search

flashlight / wav2letter

Facebook AI Research's Automatic Speech Recognition Toolkit
https://github.com/facebookresearch/wav2letter/wiki
Other
6.39k stars 1.01k forks source link

Preparing Data. #922

Closed bmblr497 closed 3 years ago

bmblr497 commented 3 years ago

Question

I want to do decoding on models present in '.../sota/2019' but i am getting stuck in first step of data preparation as I am getting the errors below.

root@d8096c5ba82c:~# python3 /root/wav2letter/recipes/models/utilities/prepare_librispeech_wp_and_official_lexicon.py --data_dst /model/audio --model_dst /model --nbest 10 --wp 10000 python3: can't open file '/root/wav2letter/recipes/models/utilities/../../../data/librispeech/prepare.py': [Errno 2] No such file or directory Preparing tokens and lexicon for acoustic model...

Traceback (most recent call last): File "/root/wav2letter/recipes/models/utilities/prepare_librispeech_wp_and_official_lexicon.py", line 80, in with open(os.path.join(lists_path, name + ".lst"), "r") as flist: FileNotFoundError: [Errno 2] No such file or directory: '/model/audio/lists/train-clean-100.lst'

Additional Context

[Add any additional information here]

abhinavkulkarni commented 3 years ago

@bmblr497: Please change lines 49-50 from

        "python3 {}/../../../data/librispeech/prepare.py --dst {} -p {}".format(
            os.path.dirname(os.path.abspath(__file__)), args.data_dst, args.process

to 

        "python3 {}/../../data/librispeech/prepare.py --dst {} -p {}".format(
            os.path.dirname(os.path.abspath(__file__)), args.data_dst, args.process
tlikhomanenko commented 3 years ago

Also please use for all code base with these pre-trained models branches v0.2 in both flashlight and wav2letter. Current master will not work with them as we changed namespacing.