---
license: cc-by-nc-4.0
---

# ECAPA2 Speaker Embedding Extractor

ECAPA2 is a hybrid neural network architecture and training strategy for speaker recognition. The provided model is pre-trained and has an easy-to-use API to extract speaker embeddings.

## Model Details

<img src="https://cdn-uploads.huggingface.co/production/uploads/620f6a7d110b521c673c1914/BORgtl2G6XUlWaZeMLGPc.png" width="300"/>

## Usage Guide

### Download model

You need to install the huggingface_hub package to download the ECAPA2 model:

```
pip install --upgrade huggingface_hub
```

Or with Conda:

```
conda install -c conda-forge huggingface_hub
```

Now you can download the model by executing the following code:

```
from huggingface_hub import hf_hub_download

model_file = hf_hub_download(repo_id=Jenthe/ECAPA2, filename='model.pt')
model = torch.jit.load(model_file, map_location='cpu')
```

Subsequent calls will load the previously downloaded model automatically.

### Speaker Embedding Extraction

Extracting speaker embeddings is easy and only requires a few lines of code:
```
import torch
import torchaudio

audio = torchaudio.load('sample.wav')
embedding = model.extract_embedding(audio)
```

### Hierarchical Feature Extraction

For the extraction of other hierachical features, a separate model function is provided:
```
feature = ecapa2_model.extract_feature(label='gfe1', type='mean')
```

The following table describes the available features:

| Feature Type| Description | Usage | Labels |
| ----------- | ----------- | ----------- | ----------- |
| Local Feature | Non-uniform effective receptive field in the frequency dimension of each frame-level feature.| Abstract features, probably usefull in tasks less related to speaker characteristics. | lfe1, lfe2, lfe3, lfe4
| Global Feature | Uniform effective receptive field of each frame-level feature in the frequency dimension.| Generally capture intra-speaker variance better then speaker embeddings. E.g. speaker profiling, emotion recognition. | gfe1, gfe2, gfe3, pool
| Speaker Embedding | Uniform effective receptive field of each frame-level feature in the frequency dimension.| Best for tasks directly depending on the speaker identity (as opposed to speaker characteristics). E.g. speaker verification, speaker diarization. | embedding


<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->

### Results

[More Information Needed]


## Citation [optional]

<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->

**BibTeX:**

```
@INPROCEEDINGS{xxxxx,
  author={Jenthe Thienpondt and Kris Demuynck},
  booktitle={2023 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)}, 
  title={ECAPA2: A Hybrid Neural Network Architecture and Training Strategy for Robust Speaker Embeddings}, 
  year={2023},
  volume={},
  number={}
}
```

**APA:**

[More Information Needed]

## Contact

Name: Jenthe Thienpondt

E-mail: jenthe.thienpondt@ugent.be