Parameter masks for close talk speech segregation using deep neural networks

Parameter masks for close talk speech segregation using deep neural networks

A deep neural networks (DNN) based close talk speech segregation algorithm is introduced. One nearby microphone is used to collect the target speech as close talk indicated, and another microphone is used to get the noise in environments. The time and energy difference between the two microphones si...

Full description

Bibliographic Details
Main Authors: Jiang Yi, Liu Runsheng
Format: Article
Language:English
Published: EDP Sciences 2015-01-01
Series:MATEC Web of Conferences
Online Access:http://dx.doi.org/10.1051/matecconf/20153117004
id doaj-11639090f1304e69b87ff3245f9ad1d6
record_format Article
spelling doaj-11639090f1304e69b87ff3245f9ad1d62021-02-02T01:46:18ZengEDP SciencesMATEC Web of Conferences2261-236X2015-01-01311700410.1051/matecconf/20153117004matecconf_icmee2015_17004Parameter masks for close talk speech segregation using deep neural networksJiang Yi0Liu Runsheng1The Quartermaster Equipment Research InstituteElectronic Engineering, Tsinghua UniversityA deep neural networks (DNN) based close talk speech segregation algorithm is introduced. One nearby microphone is used to collect the target speech as close talk indicated, and another microphone is used to get the noise in environments. The time and energy difference between the two microphones signal is used as the segregation cue. A DNN estimator on each frequency channel is used to calculate the parameter masks. The parameter masks represent the target speech energy in each time frequency (T-F) units. Experiment results show the good performance of the proposed system. The signal to noise ratio (SNR) improvement is 8.1 dB on 0 dB noisy environment.http://dx.doi.org/10.1051/matecconf/20153117004
collection DOAJ
language English
format Article
sources DOAJ
author Jiang Yi
Liu Runsheng
spellingShingle Jiang Yi
Liu Runsheng
Parameter masks for close talk speech segregation using deep neural networks
MATEC Web of Conferences
author_facet Jiang Yi
Liu Runsheng
author_sort Jiang Yi
title Parameter masks for close talk speech segregation using deep neural networks
title_short Parameter masks for close talk speech segregation using deep neural networks
title_full Parameter masks for close talk speech segregation using deep neural networks
title_fullStr Parameter masks for close talk speech segregation using deep neural networks
title_full_unstemmed Parameter masks for close talk speech segregation using deep neural networks
title_sort parameter masks for close talk speech segregation using deep neural networks
publisher EDP Sciences
series MATEC Web of Conferences
issn 2261-236X
publishDate 2015-01-01
description A deep neural networks (DNN) based close talk speech segregation algorithm is introduced. One nearby microphone is used to collect the target speech as close talk indicated, and another microphone is used to get the noise in environments. The time and energy difference between the two microphones signal is used as the segregation cue. A DNN estimator on each frequency channel is used to calculate the parameter masks. The parameter masks represent the target speech energy in each time frequency (T-F) units. Experiment results show the good performance of the proposed system. The signal to noise ratio (SNR) improvement is 8.1 dB on 0 dB noisy environment.
url http://dx.doi.org/10.1051/matecconf/20153117004
work_keys_str_mv AT jiangyi parametermasksforclosetalkspeechsegregationusingdeepneuralnetworks
AT liurunsheng parametermasksforclosetalkspeechsegregationusingdeepneuralnetworks
_version_ 1724311041840513024

Similar Items