MTASS

displayName: MTASS
license:
- MTASS Custom
mediaTypes:
- Audio
paperUrl: https://arxiv.org/pdf/2107.06467v1.pdf
publishDate: "2021"
publishUrl: https://github.com/Windstudent/Complex-MTASSNet/
publisher:
- Harbin Institute of Technology
- Kwai
tags:
- Audio track
taskTypes:
- Multi Task Audio Source Seperation

---
# 数据集介绍
## 简介
音频源分离任务，例如语音增强、语音分离和音乐源分离，在最近的研究中取得了令人瞩目的表现。深度神经网络强大的建模能力让我们对更具挑战性的任务充满希望。本文发起了一项新的多任务音频源分离 (MTASS) 挑战，将语音、音乐和噪声信号从单声道混合中分离出来。首先，我们介绍了这项任务的细节，并生成了一个包含语音、音乐和背景噪声的混合数据集。然后，我们提出了一个复杂域中的 MTASS 模型，以充分利用三个音频信号的频谱特性差异。具体来说，所提出的模型遵循两级流水线，将三种音频信号分离，然后分别进行信号补偿。在比较不同的训练目标后，选择复比掩模作为MTASS更合适的目标。实验结果还表明，残差信号补偿模块有助于进一步恢复信号。与几种众所周知的分离模型相比，所提出的模型在分离性能方面显示出显着优势。
## 引文
```
@article{zhang2021multi,
title={Multi-task audio source separation},
author={Zhang, Lu and Li, Chenxing and Deng, Feng and Wang, Xiaorui},
journal={arXiv preprint arXiv:2107.06467},
year={2021}
}
```

## Download dataset
:modelscope-code[]{type="git"}

displayName: MTASS
license:
- MTASS Custom
mediaTypes:
- Audio
paperUrl: https://arxiv.org/pdf/2107.06467v1.pdf
publishDate: "2021"
publishUrl: https://github.com/Windstudent/Complex-MTASSNet/
publisher:
- Harbin Institute of Technology
- Kwai
tags:
- Audio track
taskTypes:
- Multi-Task Audio Source Separation
---
# Dataset Introduction
## Overview
Audio source separation tasks, such as speech enhancement, speech separation, and music source separation, have achieved remarkable performance in recent research. The powerful modeling capabilities of deep neural networks have instilled hope for tackling more challenging tasks. This paper proposes a new multi-task audio source separation (MTASS) challenge, which aims to separate speech, music, and noise signals from monaural mixtures. First, we introduce the details of this task and construct a mixed dataset containing speech, music, and background noise. Then, we propose an MTASS model in the complex domain to fully leverage the spectral characteristic differences among the three audio signals. Specifically, the proposed model follows a two-stage pipeline: first separating the three audio signals, then performing individual signal compensation. After comparing different training objectives, the complex ratio mask is selected as a more suitable objective for MTASS. Experimental results also demonstrate that the residual signal compensation module aids in further signal recovery. Compared with several well-established separation models, the proposed model exhibits significant advantages in separation performance.
## Citation

@article{zhang2021multi,
title={Multi-task audio source separation},
author={Zhang, Lu and Li, Chenxing and Deng, Feng and Wang, Xiaorui},
journal={arXiv preprint arXiv:2107.06467},
year={2021}
}

## Download Dataset
:modelscope-code[]{type="git"}

搜集汇总

背景与挑战

背景概述

MTASS是一个专注于多任务音频源分离的数据集，包含语音、音乐和噪声的混合信号，支持复杂域模型的两阶段处理流程，实验结果显示其分离性能优越。