Skip to content

nikola-j/audio_tag

Repository files navigation

Audiotag

Automatic .wav file audio tagging using VGGish. A simple POC.

Installation

Install requirements using:

pip install -r requiremnts.txt

Download VGGish model:

# Download data files into the audioset directory
cd audioset
curl -O https://storage.googleapis.com/audioset/vggish_model.ckpt
curl -O https://storage.googleapis.com/audioset/vggish_pca_params.npz

Dataset

Download dataset from this competition: https://www.kaggle.com/c/freesound-audio-tagging

Pre-compute vggish audio features

Use batch_inference.py to compute audio features from audio signals eg:

python batch_inference.py --wav_train [train files] --wav_csv [train csv file]

Train

Use train.py to train.py a model, choose a batch size and model to use

Inference

Use jupyter to run inference, open 'Sound tag.ipynb'

External code

Audioset used from here: https://github.com/tensorflow/models

Batched vggish inference used from here: https://github.com/knstmrd/vggish-batch

Releases

No releases published

Packages

No packages published