audio_classification
audio_classification copied to clipboard
Audio classification using CNN and LSTM
Data Visualization
MFCC Features
![](https://github.com/geekysethi/audio_classification/raw/master/images/mfcc.png)
Spectrogram
![](https://github.com/geekysethi/audio_classification/raw/master/images/spec.png)
Raw Audio
![](https://github.com/geekysethi/audio_classification/raw/master/images/raw.png)
Results
CNN | ||||
---|---|---|---|---|
Spectrogram | MFCC | |||
Dataset | Train | Validation | Train | Validation |
urbansound8k | 99.914 | 97.252 | 100 | 84.544 |
CNN-LSTM | ||||
---|---|---|---|---|
Spectrogram | MFCC | |||
Dataset | Train | Validation | Train | Validation |
urbansound8k | 99.928 | 96.451 | 99.985 | 82.369 |
Training Accuracy Plot
![](https://github.com/geekysethi/audio_classification/raw/master/images/Section-0-Panel-0-3919lbfls.png)
Validation Accuracy Plot
![](https://github.com/geekysethi/audio_classification/raw/master/images/Section-0-Panel-1-jmn9k1muq.png)
Training Error Plot
![](https://github.com/geekysethi/audio_classification/raw/master/images/Section-0-Panel-2-99q67v5kr.png)
Validation Error Plot
![](https://github.com/geekysethi/audio_classification/raw/master/images/Section-0-Panel-3-tg061apfa.png)
Installation
Use the package manager pip to install foobar.
pip install requirements.txt
or
conda create --name <env> --file requirements.txt
Usage
Dataset
Pre-process Data
python codes/pre_processing/pre_processing_urbansound.py
Train and Test
python codes/baseline/main.py
Contributing
Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.
Please make sure to update tests as appropriate.