Spectral subtraction using correct convolution and a spectrum dependent exponential averaging method.

Document type: Researchreports
Full text:
Author(s): Harald Gustafsson, Sven Nordholm, Ingvar Claesson
Title: Spectral subtraction using correct convolution and a spectrum dependent exponential averaging method.
Translated title: Spektral subtraktion med korrekt faltning samt en spektrum beroende exponentiell medelvärdesbildningsmetod.
Series: Research Report
Year: 1998
Issue: 15
ISSN: 1103-1581
Organization: Blekinge Institute of Technology
Department: Dept. of Signal Processing (Institutionen för signalbehandling)
Dept. of Signal Processing S-372 25 Ronneby
+46 455 780 00
http://www.hk-r.se/isb/isb_en.html
Authors e-mail: Harald.Gustafsson@isb.hk-r.se
Language: English
Abstract: In handsfree speech communication the signal to noise ratio is poor which makes it difficult for the listener to have a relaxed conversation. By using speech enhancement processing, the quality of conversation will be enhanced. This paper describes a speech enhancement algorithm
based on spectral subtraction. The method employs a noise and speech dependent gain function which is used to design a filter. The proper measures have been taken to obtain a causal filter and also to ensure that the circular convolution yields a linear filtering. A novel method that uses spectrum-dependent exponential averaging to decrease the variance of the gain function is also presented. The result obtained is a 13 to 18 dB noise reduction, with minor speech distortion and moderate residual noise distortion.
Summary in Swedish: Vid handsfree röstkommunikation i bullriga miljöer är det svårt att ha en avslappnad konversation. För att förbättra kommunikations möjligheterna kan man applicera brusredusering på mikrofonsignalen. Denna forskningsrapport beskriver en brusreduseringsmetod som använder en vidareutvecklad Spektral subtraktionsmetod. Vidare utvecklingen består av att filtrering görs med korrekt faltning samt en spektrum beroende exponentiell medelvärdesbildningsmetod.
Subject: Signal Processing\Speech Enhancement
Keywords: speech enhancement, noise reduction, spectral subtraction
URN: urn:nbn:se:bth-00123
Edit