Name: sepformer wham16k enhancement
Author: speechbrain

Overview

The SepFormer WHAM!16k is a specialized speech enhancement model designed to isolate clean voice signals from noisy environments. Unlike traditional spectral subtraction methods, it leverages a transformer-based architecture to handle long-range dependencies in audio, making it particularly effective at removing non-stationary background noise while preserving speech intelligibility. For developers, this model is ideal for preprocessing pipelines in ASR (Automatic Speech Recognition) or VoIP applications where audio quality is inconsistent. It integrates seamlessly via the SpeechBrain framework, offering a robust alternative to simpler RNN-based denoisers by providing superior separation performance on the WHAM! dataset benchmarks.

Highlights

Transformer-based architecture for superior non-stationary noise removal
Optimized for 16kHz audio sampling rates
Seamless integration via the SpeechBrain toolkit
Apache-2.0 license for flexible commercial deployment
Ideal preprocessing for improving ASR accuracy

Usage

Install

# Install Hugging Face transformers
pip install transformers torch

SDK Usage

# Load model with transformers
from transformers import AutoModel, AutoTokenizer

model = AutoModel.from_pretrained("speechbrain/sepformer-wham16k-enhancement")
tokenizer = AutoTokenizer.from_pretrained("speechbrain/sepformer-wham16k-enhancement")

Hugging Face Download

We recommend downloading the model via the Hugging Face CLI or Hub SDK.

Guidance：Before downloading, install huggingface_hub with:

Guidance

pip install -U huggingface_hub

CLI Download

Download the full repository

huggingface-cli download speechbrain/sepformer-wham16k-enhancement

Download a single file to a local folder (e.g. config.json into ./dir)

huggingface-cli download speechbrain/sepformer-wham16k-enhancement config.json --local-dir ./dir

See the official docs for more CLI options

SDK Download

# 模型下载
from huggingface_hub import snapshot_download
model_dir = snapshot_download('speechbrain/sepformer-wham16k-enhancement')

Git Download

Make sure git-lfs is installed first

Git Download

git lfs install
git clone https://huggingface.co/speechbrain/sepformer-wham16k-enhancement

To skip LFS large-file downloads, use:

Skip LFS

GIT_LFS_SKIP_SMUDGE=1 git clone https://huggingface.co/speechbrain/sepformer-wham16k-enhancement

Model files are hosted on the Hugging Face Hub — download directly via HF CLI / SDK / Git, not through this site.

PyTorch / Transformers Usage

Install Transformers

pip install -U transformers torch

Load the model and run inference

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained('speechbrain/sepformer-wham16k-enhancement')
tokenizer = AutoTokenizer.from_pretrained('speechbrain/sepformer-wham16k-enhancement')

Model Download

We recommend downloading the model via the ModelScope CLI or SDK.

Guidance：Before downloading, install ModelScope with:

Guidance

pip install modelscope

CLI Download

Download the full repository

modelscope download --model speechbrain/sepformer-wham16k-enhancement

Download a single file to a local folder (e.g. README.md into ./dir)

modelscope download --model speechbrain/sepformer-wham16k-enhancement README.md --local_dir ./dir

See the docs for more CLI options

SDK Download

# 模型下载
from modelscope import snapshot_download
model_dir = snapshot_download('speechbrain/sepformer-wham16k-enhancement')

Git Download

Make sure git-lfs is installed first

Git Download

git lfs install
git clone https://www.modelscope.cn/speechbrain/sepformer-wham16k-enhancement.git

To skip LFS large-file downloads, use:

Skip LFS

GIT_LFS_SKIP_SMUDGE=1 git clone https://www.modelscope.cn/speechbrain/sepformer-wham16k-enhancement.git

ModelScope 模型页直接下载模型文件；无需将模型文件放在本站服务器。

Notebook Quickstart

Install the ModelScope library

pip install "modelscope[audio,cv,nlp,multi-modal,science]" -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html

Load the model and run inference

from modelscope.pipelines import pipeline
from modelscope.utils.constant import Tasks

p = pipeline('text-generation', 'speechbrain/sepformer-wham16k-enhancement')

Full Documentation

来源: HuggingFace

---
language: "en"
thumbnail:
tags:

audio-to-audio

Speech Enhancement

WHAM!

SepFormer

Transformer

pytorch

speechbrain

license: "apache-2.0"
datasets:

WHAM

metrics:

SI-SNR

PESQ

---

SepFormer trained on WHAM! for speech enhancement (16k sampling frequency)

This repository provides all the necessary tools to perform speech enhancement (denoising) with a SepFormer model, implemented with SpeechBrain, and pretrained on WHAM! dataset with 16k sampling frequency, which is basically a version of WSJ0-Mix dataset with environmental noise and reverberation in 8k. For a better experience we encourage you to learn more about SpeechBrain. The given model performance is 14.3 dB SI-SNR on the test set of WHAM! dataset.

| Release | Test-Set SI-SNR | Test-Set PESQ |
|:-------------:|:--------------:|:--------------:|
| 06-30-22 | 13.8 | 2.20 |

Install SpeechBrain

First of all, please install SpeechBrain with the following command:

code

pip install speechbrain

Please notice that we encourage you to read our tutorials and learn more about SpeechBrain.

Perform speech enhancement on your own audio file

python

from speechbrain.pretrained import SepformerSeparation as separator
import torchaudio
model = separator.from_hparams(source="speechbrain/sepformer-wham16k-enhancement", savedir='pretrained_models/sepformer-wham16k-enhancement')
for custom file, change path
est_sources = model.separate_file(path='speechbrain/sepformer-wham16k-enhancement/example_wham16k.wav')
torchaudio.save("enhanced_wham16k.wav", est_sources[:, :, 0].detach().cpu(), 16000)

Inference on GPU

To perform inference on the GPU, add run_opts={"device":"cuda"} when calling the from_hparams method.

Training

The training script is currently being worked on an ongoing pull-request.

We will update the model card as soon as the PR is merged.

You can find our training results (models, logs, etc) here.

Limitations

The SpeechBrain team does not provide any warranty on the performance achieved by this model when used on other datasets.

#### Referencing SpeechBrain

bibtex

@misc{speechbrain,
  title={{SpeechBrain}: A General-Purpose Speech Toolkit},
  author={Mirco Ravanelli and Titouan Parcollet and Peter Plantinga and Aku Rouhe and Samuele Cornell and Loren Lugosch and Cem Subakan and Nauman Dawalatabad and Abdelwahab Heba and Jianyuan Zhong and Ju-Chieh Chou and Sung-Lin Yeh and Szu-Wei Fu and Chien-Feng Liao and Elena Rastorgueva and François Grondin and William Aris and Hwidong Na and Yan Gao and Renato De Mori and Yoshua Bengio},
  year={2021},
  eprint={2106.04624},
  archivePrefix={arXiv},
  primaryClass={eess.AS},
  note={arXiv:2106.04624}
}

#### Referencing SepFormer

bibtex

@inproceedings{subakan2021attention,
      title={Attention is All You Need in Speech Separation}, 
      author={Cem Subakan and Mirco Ravanelli and Samuele Cornell and Mirko Bronzi and Jianyuan Zhong},
      year={2021},
      booktitle={ICASSP 2021}
}
@article{subakan2023exploring,
  author={Subakan, Cem and Ravanelli, Mirco and Cornell, Samuele and Grondin, François and Bronzi, Mirko},
  journal={IEEE/ACM Transactions on Audio, Speech, and Language Processing}, 
  title={Exploring Self-Attention Mechanisms for Speech Separation}, 
  year={2023},
  volume={31},
  pages={2169-2180},
}

About SpeechBrain

Website: https://speechbrain.github.io/

Code: https://github.com/speechbrain/speechbrain/

HuggingFace: https://huggingface.co/speechbrain/