search

hirofumi0810 / asr_preprocessing

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

location of make.sh #3

Closed cuhkebook closed 6 years ago

cuhkebook commented 6 years ago

May I know where can find the make.sh in order to set the path? Many thanks!

hirofumi0810 commented 6 years ago

@cuhkebook

make.sh was replaced with data_prep.sh. I will update README.md.

Thank you.

cuhkebook commented 6 years ago

Thank a lot! I found it. After update the data_prep.sh and run ./data_prep.sh

Traceback (most recent call last): File "make_config.py", line 15, in from timit.path import Path ImportError: No module named timit.path

Where can I find the timit.path in order to update it? Thank you!

hirofumi0810 commented 6 years ago

@cuhkebook I think it is due to python2. Therefore, please add __init__.py.

cuhkebook commented 6 years ago

O...thank a lot! It is solved as Add "init.py" into timit folder, which is an empty directory

Also need one more dependence: pip install librosa

After, try again the ./data_prep.sh and get it:

===> Reading audio files... 0it [00:00, ?it/s] ===> Computing global mean & std over the training set... Traceback (most recent call last): File "main.py", line 165, in main() File "main.py", line 160, in main make_input() File "main.py", line 88, in make_input save_path=mkdir_join(input_save_path, 'train')) File "../timit/input_data.py", line 151, in read_audio feat_dim = input_data_list_male[0].shape[1] IndexError: list index out of range

I figure out that my TIMIT corpus is in capital letter of "SX", "SI" and ".WAV", so the path.py cannot detect the filename.

hirofumi0810 commented 6 years ago

@cuhkebook I think this is due to wrong path to TIMIT corpus. Pleach check path.