https://github.com/zhaoyi2/audio_augment
A tool/script for batch speech data enhancement with speed/volume/RIRS/MUSAN
https://github.com/zhaoyi2/audio_augment
audio-augmentation musan optional rirs speed volume
Last synced: 6 months ago
JSON representation
A tool/script for batch speech data enhancement with speed/volume/RIRS/MUSAN
- Host: GitHub
- URL: https://github.com/zhaoyi2/audio_augment
- Owner: zhaoyi2
- Created: 2020-06-27T15:18:37.000Z (over 5 years ago)
- Default Branch: master
- Last Pushed: 2020-06-28T10:25:05.000Z (over 5 years ago)
- Last Synced: 2024-12-13T01:34:32.055Z (about 1 year ago)
- Topics: audio-augmentation, musan, optional, rirs, speed, volume
- Language: Shell
- Homepage:
- Size: 2.39 MB
- Stars: 22
- Watchers: 1
- Forks: 4
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Audio augmentation
(README.md)
*Audio augment* is a tool/script for batch audio data augmenation through speed, volume, reverb, noise based on kaldi and sox.
## Installation
- kaldi
- sox
- cd audio_augment
- cd tools; make KALDI=/your_kaldi_path
## Usage
- cd audio_augment
- vim run_aug.sh to change your input_path and out_path and save
- bash run_aug.sh, so easy!
## Workflow(run_aug.sh)
- Stage 1: Data Preparation contain data and text
- Stage 2: speed 0.9/1.1
- Stage 3: volume +-db
- Stage 4: reverberation(RIRS)
- Stage 5: MUSAN(noise/music/babble)
- Stage 6: combine above data and select a subset of the augmend data list about twice the origin data
- Stage 7: data and label generation
## generated Examples
- cd data/wav/train_aug listen a few enhanced aishell1 audio example through speed/volume/RIRS/MUSAN.
### Reference
- MUSAN http://www.openslr.org/17/
- RIRS http://www.openslr.org/28/
- https://github.com/linan2/add_reverb2