search

hirofumi0810 / asr_preprocessing

Python implementation of pre-processing for End-to-End speech recognition
MIT License
69 stars 23 forks source link

couple of easy to fix bugs #2

Closed jesuistay closed 6 years ago

jesuistay commented 6 years ago

(1) In wav2feature_librosa.py and wav2feature_python_speech_features.py delta1 does not get initialized if use_delta1 is true and use_delta2 is false.

(2) When preprocessing only libri-speech-100 a character.txt file does not get created. My workaround was that I preprocessed train_all but fixed the paths so only the ones from libri-speech-100 would get used.

hirofumi0810 commented 6 years ago

@jesuistay , thank you. I'll fix them in a few days. After that, you will be able to create dataset only with 100h data (train-clean-100).

Please give me a moment.

jesuistay commented 6 years ago

Great to hear. After a couple of days with the project, I gotta commend you for a great piece of code overall. Well earned star.