Noise-mixer is an useful data generate tool writen by Python 2.7, which can mix noise wav to your origin clear wav file.
By generating noisy audio file, Noise-mixer can help you get more training data when train an ASR(automatic speech recognition) system.
It's easy to generate audio file(wav file) by noise-mixer, what you need to do is:
- Prepare noise wav list, which including all noise wav in this format:
name path
sample like:
- Perpare origin clear wav list in this format:
name path
sample like:
- Edit
produce.sh
to change your output directory, noise_home, wav_home and noise level and so on. - Then run
produce.sh
, wait a moment for the output. - Done !
This tool is developed by Python 2.7, it doesn't work for Python 3.*