HEAR 2021 NeurIPS Challenge: Wav2CLIP

https://neuralaudio.ai/hear2021-holistic-evaluation-of-audio-representations.html

Installation

pip install git+https://github.com/hohsiangwu/wav2clip-hear.git

Usage

Scene Embeddings

import wav2clip_hear

model = wav2clip_hear.load_model("")
embedding = wav2clip_hear.get_scene_embeddings(audio, model)

Timestamp Embeddings

import wav2clip_hear

model = wav2clip_hear.load_model("")
embedding, timestamp = wav2clip_hear.get_timestamp_embeddings(audio, model)

Model Weights License

The model weights are made available under a Creative Commons Attribution 4.0 International (CC BY 4.0) License.