Skip to main content

CLAP (Contrastive Language-Audio Pretraining) is a model that learns acoustic concepts from natural language supervision and enables “Zero-Shot” inference. The model has been extensively evaluated in 26 audio downstream tasks achieving SoTA in several of them including classification, retrieval, and captioning.

Project description

Overview | Setup | CLAP weights | Usage | Examples | Citation

CLAP

CLAP (Contrastive Language-Audio Pretraining) is a model that learns acoustic concepts from natural language supervision and enables “Zero-Shot” inference. The model has been extensively evaluated in 26 audio downstream tasks achieving SoTA in several of them including classification, retrieval, and captioning.

clap_diagrams

Setup

First, install python 3.8 or higher (3.11 recommended). Then, install CLAP using either of the following:

# Install pypi pacakge
pip install msclap

# Or Install latest (unstable) git source
pip install git+https://github.com/microsoft/CLAP.git

CLAP weights

CLAP weights are downloaded automatically (choose between versions 2022, 2023, and clapcap), but are also available at: Zenodo or HuggingFace

clapcap is the audio captioning model that uses the 2023 encoders.

Usage

  • Zero-Shot Classification and Retrieval
from msclap import CLAP

# Load model (Choose between versions '2022' or '2023')
# The model weight will be downloaded automatically if `model_fp` is not specified
clap_model = CLAP(version = '2023', use_cuda=False)

# Extract text embeddings
text_embeddings = clap_model.get_text_embeddings(class_labels: List[str])

# Extract audio embeddings
audio_embeddings = clap_model.get_audio_embeddings(file_paths: List[str])

# Compute similarity between audio and text embeddings 
similarities = clap_model.compute_similarity(audio_embeddings, text_embeddings)
  • Audio Captioning
from msclap import CLAP

# Load model (Choose version 'clapcap')
clap_model = CLAP(version = 'clapcap', use_cuda=False)

# Generate audio captions
captions = clap_model.generate_caption(file_paths: List[str])

Examples

Take a look at examples for usage examples.

To run Zero-Shot Classification on the ESC50 dataset try the following:

> cd examples && python zero_shot_classification.py

Output (version 2023)

ESC50 Accuracy: 93.9%

Citation

Kindly cite our work if you find it useful.

CLAP: Learning Audio Concepts from Natural Language Supervision

@inproceedings{CLAP2022,
  title={Clap learning audio concepts from natural language supervision},
  author={Elizalde, Benjamin and Deshmukh, Soham and Al Ismail, Mahmoud and Wang, Huaming},
  booktitle={ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  pages={1--5},
  year={2023},
  organization={IEEE}
}

Natural Language Supervision for General-Purpose Audio Representations

@misc{CLAP2023,
      title={Natural Language Supervision for General-Purpose Audio Representations}, 
      author={Benjamin Elizalde and Soham Deshmukh and Huaming Wang},
      year={2023},
      eprint={2309.05767},
      archivePrefix={arXiv},
      primaryClass={cs.SD},
      url={https://arxiv.org/abs/2309.05767}
}

Contributing

This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.opensource.microsoft.com.

When you submit a pull request, a CLA bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., status check, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.

This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact opencode@microsoft.com with any additional questions or comments.

Trademarks

This project may contain trademarks or logos for projects, products, or services. Authorized use of Microsoft trademarks or logos is subject to and must follow Microsoft's Trademark & Brand Guidelines. Use of Microsoft trademarks or logos in modified versions of this project must not cause confusion or imply Microsoft sponsorship. Any use of third-party trademarks or logos are subject to those third-party's policies.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

msclap-1.3.3.tar.gz (27.5 kB view details)

Uploaded Source

Built Distribution

msclap-1.3.3-py3-none-any.whl (31.4 kB view details)

Uploaded Python 3

File details

Details for the file msclap-1.3.3.tar.gz.

File metadata

  • Download URL: msclap-1.3.3.tar.gz
  • Upload date:
  • Size: 27.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/1.6.1 CPython/3.8.5 Windows/10

File hashes

Hashes for msclap-1.3.3.tar.gz
Algorithm Hash digest
SHA256 ded253972b745d4e06026dda408dc80f625717e6887837ce4e6ec3e633052269
MD5 9258a4c87ee49acf0783e187b3976882
BLAKE2b-256 b01b6605192e5f4ccf9eef13aaea46443e0658db61f11abb8910373b13089926

See more details on using hashes here.

File details

Details for the file msclap-1.3.3-py3-none-any.whl.

File metadata

  • Download URL: msclap-1.3.3-py3-none-any.whl
  • Upload date:
  • Size: 31.4 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/1.6.1 CPython/3.8.5 Windows/10

File hashes

Hashes for msclap-1.3.3-py3-none-any.whl
Algorithm Hash digest
SHA256 ee9f36ecbb19d8cc4dee42c54a1b48d9fae2b01c9b4532ef35be18533b1ea8c3
MD5 cfc7ea4b1c54bfeb29ec179bf6afc1d1
BLAKE2b-256 28aec22b7e05ec7864637d39acede8bdc5e36a481b9701ad47ed7e754ad5155d

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page