Closed canhbd closed 6 months ago
Hello,
I'm reaching out to inquire if you've cloned the dataset from SpeechOcean762 (SO762). It seems that the GitHub repository for SO762 has been updated, and the .wav files there might not be the original ones I've been using.
Although I haven't updated the introduction in my README yet, I want to inform you that there is download check for SO762 in the latest version of gen_seq_data_utt.py
. So, try to delete the original dataset that you clone and run the code again. The new downloaded dictionary should function properly in the prep_data process.
BTW, the download check shows below:
dataset_path = '../speechocean762'
if not os.path.exists(dataset_path):
# change to the directory where you want to save the dataset
os.chdir('../')
# download the dataset
os.system('wget https://www.openslr.org/resources/101/speechocean762.tar.gz')
os.system('tar -xvzf speechocean762.tar.gz')
os.chdir('prep_data')
I hope this information helps you.
Hello, I don't see prep/data folder, I just only see prep_data folder
you mean create new folder?
I have tried many times (both: prep_data and create new folder), but still error :(
Sorry, it should be prep_data
. It's a typo and the command is the part of run.sh
. (my bad, there are misleading directions....)
In original repository, there are a prep_data
folder and the command you may need are in run.sh
.
Sorry again for the mistake, and I will update README.
Hello, Finally, I can run it. I have missed version of torch (maybe, im not familiar with python. So, the system install version 2.2 as default) Thank for your careful support 💯
I will close this issue now Thank you!
Hello, the first thank for your sharing this repo.
I have a problem when run this, this is error message :
_RuntimeError: Couldn't find appropriate backend to handle uri ../speechocean762/WAVE/SPEAKER0001/000010011.WAV and format None._
This is my terminal:
This is my folder structure:
How can I fix it? Thank you!