Wespeaker/wespeaker-ecapa-tdnn512-LM

Official model provided by Wespeaker project, ECAPA-TDNN based x-vector (After large margin finetune)

The model is trained on VoxCeleb2 Dev dataset, containing 5994 speakers.

Model Sources

• Repository: https://github.com/wenet-e2e/wespeaker
• Paper: https://arxiv.org/pdf/2210.17016.pdf
• Demo: https://huggingface.co/spaces/wenet/wespeaker_demo

Results on VoxCeleb

Model	Params	Flops	LM	AS-Norm	vox1-O-clean	vox1-E-clean	vox1-H-clean
ECAPA_TDNN_GLOB_c512-ASTP-emb192	6.19M	1.04G	×	×	1.069	1.209	2.310
			×	√	0.957	1.128	2.105
			√	×	0.878	1.072	2.007
			√	√	0.782	1.005	1.824

Install Wespeaker

pip install git+https://github.com/wenet-e2e/wespeaker.git

for development install:

git clone https://github.com/wenet-e2e/wespeaker.git
cd wespeaker
pip install -e .

Command line Usage

$ wespeaker -p ecapa_tdnn512_download_dir --task embedding --audio_file audio.wav --output_file embedding.txt
$ wespeaker -p ecapa_tdnn512_download_dir --task embedding_kaldi --wav_scp wav.scp --output_file /path/to/embedding
$ wespeaker -p ecapa_tdnn512_download_dir --task similarity --audio_file audio.wav --audio_file2 audio2.wav
$ wespeaker -p ecapa_tdnn512_download_dir --task diarization --audio_file audio.wav

Python Programming Usage

import wespeaker

model = wespeaker.load_model_local(ecapa_tdnn512_download_dir)
# set_gpu to enable the cuda inference, number < 0 means using CPU
model.set_gpu(0)

# embedding/embedding_kaldi/similarity/diarization
embedding = model.extract_embedding('audio.wav')
utt_names, embeddings = model.extract_embedding_list('wav.scp')
similarity = model.compute_similarity('audio1.wav', 'audio2.wav')
diar_result = model.diarize('audio.wav')

# register and recognize
model.register('spk1', 'spk1_audio1.wav')
model.register('spk2', 'spk2_audio1.wav')
model.register('spk3', 'spk3_audio1.wav')
result = model.recognize('spk1_audio2.wav')

Citation

@article{desplanques2020ecapa,
  title={Ecapa-tdnn: Emphasized channel attention, propagation and aggregation in tdnn based speaker verification},
  author={Desplanques, Brecht and Thienpondt, Jenthe and Demuynck, Kris},
  journal={arXiv preprint arXiv:2005.07143},
  year={2020}
}

@inproceedings{wang2023wespeaker,
  title={Wespeaker: A research and production oriented speaker embedding learning toolkit},
  author={Wang, Hongji and Liang, Chengdong and Wang, Shuai and Chen, Zhengyang and Zhang, Binbin and Xiang, Xu and Deng, Yanlei and Qian, Yanmin},
  booktitle={IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  pages={1--5},
  year={2023},
  organization={IEEE}
}