DeepMMSE: A Deep Learning Approach to MMSE-Based Noise Power Spectral Density Estimation

噪音(视频) 计算机科学 估计员 噪声功率 噪声测量 噪声地板 数值噪声 光谱密度 梯度噪声 语音增强 最小均方误差 语音识别 人工智能 算法 功率(物理) 数学 降噪 统计 电信 物理 图像(数学) 量子力学
作者
Qiquan Zhang,Aaron Nicolson,Mingjiang Wang,Kuldip K. Paliwal,C.P. Wang
出处
期刊:IEEE/ACM transactions on audio, speech, and language processing [Institute of Electrical and Electronics Engineers]
卷期号:28: 1404-1415 被引量:119
标识
DOI:10.1109/taslp.2020.2987441
摘要

An accurate noise power spectral density (PSD) tracker is an indispensable component of a single-channel speech enhancement system. Bayesian-motivated minimum mean-square error (MMSE)-based noise PSD estimators have been the most prominent in recent time. However, they lack the ability to track highly non-stationary noise sources due to current methods of a priori signal-to-noise (SNR) estimation. This is caused by the underlying assumption that the noise signal changes at a slower rate than the speech signal. As a result, MMSE-based noise PSD trackers exhibit a large tracking delay and produce noise PSD estimates that require bias compensation. Motivated by this, we propose an MMSE-based noise PSD tracker that employs a temporal convolutional network (TCN) a priori SNR estimator. The proposed noise PSD tracker, called DeepMMSE makes no assumptions about the characteristics of the noise or the speech, exhibits no tracking delay, and produces an accurate estimate that requires no bias correction. Our extensive experimental investigation shows that the proposed DeepMMSE method outperforms state-of-the-art noise PSD trackers and demonstrates the ability to track abrupt changes in the noise level. Furthermore, when employed in a speech enhancement framework, the proposed DeepMMSE method is able to outperform state-of-the-art noise PSD trackers, as well as multiple deep learning approaches to speech enhancement. Availability: DeepMMSE is available at: https://github.com/anicolson/DeepXi.
最长约 10秒,即可获得该文献文件

科研通智能强力驱动
Strongly Powered by AbleSci AI
更新
大幅提高文件上传限制,最高150M (2024-4-1)

科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
Leo发布了新的文献求助10
1秒前
小巧蛋挞完成签到,获得积分10
1秒前
1秒前
lao333完成签到,获得积分10
1秒前
科研通AI2S应助与我常在采纳,获得10
2秒前
Akim应助jewelliang采纳,获得10
2秒前
shan发布了新的文献求助10
2秒前
chen完成签到,获得积分10
4秒前
隐形曼青应助科研通管家采纳,获得10
4秒前
Lucas应助科研通管家采纳,获得10
4秒前
拼搏笑阳应助科研通管家采纳,获得10
5秒前
脑洞疼应助科研通管家采纳,获得10
5秒前
赘婿应助科研通管家采纳,获得30
5秒前
Vxfhfdhkcds完成签到 ,获得积分10
5秒前
wanci应助科研通管家采纳,获得10
5秒前
李爱国应助科研通管家采纳,获得10
5秒前
研友_VZG7GZ应助科研通管家采纳,获得10
5秒前
orixero应助科研通管家采纳,获得10
5秒前
淡定的松子完成签到,获得积分10
5秒前
5秒前
1+1应助科研通管家采纳,获得10
5秒前
我是老大应助科研通管家采纳,获得10
5秒前
Akim应助科研通管家采纳,获得10
5秒前
拼搏笑阳应助科研通管家采纳,获得10
5秒前
今后应助科研通管家采纳,获得10
5秒前
传奇3应助科研通管家采纳,获得10
5秒前
一只饺子应助科研通管家采纳,获得30
5秒前
所所应助科研通管家采纳,获得10
5秒前
JamesPei应助科研通管家采纳,获得10
5秒前
6秒前
6秒前
李桑尼发布了新的文献求助30
6秒前
萝卜不是菠萝完成签到,获得积分10
7秒前
ding应助小兜豆豆采纳,获得10
7秒前
renpp完成签到,获得积分10
7秒前
Hello应助輝23采纳,获得10
7秒前
乐乐应助忧郁的依珊采纳,获得10
8秒前
masonzhang发布了新的文献求助10
8秒前
标致惋庭完成签到,获得积分20
9秒前
不喜欢玩球完成签到,获得积分10
9秒前
高分求助中
Earth System Geophysics 1000
Studies on the inheritance of some characters in rice Oryza sativa L 600
Medicina di laboratorio. Logica e patologia clinica 600
mTOR signalling in RPGR-associated Retinitis Pigmentosa 500
A new species of Velataspis (Hemiptera Coccoidea Diaspididae) from tea in Assam 500
Aspects of Babylonian celestial divination: the lunar eclipse tablets of Enūma Anu Enlil 500
Semiconductor Process Reliability in Practice 500
热门求助领域 (近24小时)
化学 医学 生物 材料科学 工程类 有机化学 生物化学 物理 内科学 纳米技术 计算机科学 化学工程 复合材料 基因 遗传学 催化作用 物理化学 免疫学 量子力学 细胞生物学
热门帖子
关注 科研通微信公众号,转发送积分 3206078
求助须知:如何正确求助?哪些是违规求助? 2855402
关于积分的说明 8099348
捐赠科研通 2520496
什么是DOI,文献DOI怎么找? 1353397
科研通“疑难数据库(出版商)”最低求助积分说明 641741
邀请新用户注册赠送积分活动 612821