AES London 2010
Paper Session P12
P12 - Noise Reduction and Speech Intelligibility
Sunday, May 23, 14:00 — 17:30 (Room C5)
Chair: Rhonda Wilson, Meridian Audio - Huntingdon, UK
P12-1 Monolateral and Bilateral Fitting with Different Hearing Aids Directional Configurations—Lorenzo Picinali, De Montfort University - Leicester, UK; Silvano Prosser, Università di Ferrara - Ferrara, Italy
Hearing aid bilateral fitting in hearing impaired subjects raises some problems concerning the interaction of the perceptual binaural properties with the directional characteristics of the device. This experiment aims to establish whether and to which extent in a sample of 20 normally hearing subjects the binaural changes in the speech-to-noise level ratio (s/n), caused by different symmetrical and asymmetrical microphone configurations, and different positions of the speech signal (frontal or lateral), could alter the performances of the speech recognition in noise. Speech Reception Thresholds (SRT) in noise (simulated through an Ambisonic virtual sound field) have been measured monolaterally and bilaterally in order to properly investigate the role of the binaural interaction in the perception of reproduced signals.
Convention Paper 8046 (Purchase now)
P12-2 Acoustic Echo Cancellation for Wideband Audio and Beyond—Shreyas Paranjpe, Scott Pennock, Phil Hetherington, QNX Software Systems Inc. (Wavemakers) - Vancouver, BC, Canada
Speech processing is finally starting the transition to wider bandwidths. The benefits include increased intelligibility and comprehension and a more pleasing communication experience. High quality full-duplex Acoustic Echo Cancellation (AEC) is an integral component of a hands-free speakerphone communication system because it allows participants to converse in a natural manner (and without a headset!) as they would in person. Some high-end Telepresence systems already achieve life-like communication but are computationally demanding and prohibitively expensive. The challenge is to develop a robust Acoustic Echo Canceler (AEC) that processes full-band audio signals with low computational complexity and reasonable memory consumption for an affordable Telepresence experience.
Convention Paper 8047 (Purchase now)
P12-3 Adaptive Noise Reduction for Real-Time Applications—Constantin Wiesener, TU Berlin - Berlin, Germany; Tim Flohrer, Alexander Lerch, zplane.development - Berlin, Germany; Stefan Weinzierl, TU Berlin - Berlin, Germany
We present a new algorithm for real-time noise reduction of audio signals. In order to derive the noise reduction function, the proposed method adaptively estimates the instantaneous noise spectrum from an autoregressive signal model as opposed to the widely-used approach of using a constant noise spectrum fingerprint. In conjunction with the Ephraim and Malah suppression rule a significant reduction of both stationary and non-stationary noise can be obtained. The adaptive algorithm is able to work without user interaction and is capable of real-time processing. Furthermore, quality improvements are easily possible by integration of additional processing blocks such as transient preservation.
Convention Paper 8048 (Purchase now)
P12-4 Active Noise Reduction in Personal Audio Delivery Systems; Assessment Using Loudness Balance Methods—Paul Darlington, Pierre Guiu, Phitek Systems (Europe) Sarl - Lausanne, Switzerland
Subjective methods developed for rating passive hearing protectors are inappropriate to measure active noise reduction of earphones or headphones. Additionally, assessment using objective means may produce misleading results, which do not correlate well with wearer experience and do not encompass human variability. The present paper describes application of “loudness balance” methods to the estimation of active attenuation of consumer headphones and earphones. Early results suggest that objective measures of the active reduction of pressure report greater attenuations than those implied by estimates of perceived loudness, particularly at low frequency.
Convention Paper 8049 (Purchase now)
P12-5 Mapping Speech Intelligibility in Noisy Rooms—John F. Culling, Sam Jelfs, Cardiff University - Cardiff, UK; Mathieu Lavandier, Université de Lyon - Vauix-en-Velin Cedex, France
We have developed an algorithm for accurately predicting the intelligibility of speech in noise in a reverberant environment. The algorithm is based on a development of the equalization-cancellation theory of binaural unmasking, combined with established prediction methods for monaural speech perception in noise. It has been validated against a wide range of empirical data. Acoustic measurements of rooms, known as binaural room impulse responses (BRIRs) are analyzed to predict intelligibility of a nearby voice masked by any number of steady-state noise maskers in any spatial configuration within the room. This computationally efficient method can be used to generate intelligibility maps of rooms based on the design of the room.
Convention Paper 8050 (Purchase now)
P12-6 Further Investigations into Improving STI’s Recognition of the Effects of Poor Frequency Response on Subjective Intelligibility—Glenn Leembruggen, Acoustic Directions Pty Ltd. - ICE Design, Sydney, Australia, and University of Sydney, Sydney, Australia; Marco Hippler, University of Applied Sciences Cologne - Cologne, Germany; Peter Mapp, Peter Mapp and Associates - Colchester, UK
Previous work has highlighted deficiencies in the ability of the STI metric to satisfactorily recognize the subjective loss of intelligibility that occurs with sound systems having poor frequency responses, particularly in the presence of reverberation. In a recent paper we explored the changes to STI values resulting from a range of dynamic speech spectra taken over differing time lengths with different filter responses. That work included determining the effects on STI values of three alternative spreading functions simulating the ear’s upward masking mechanism. This paper extends that work and explores the effects on STI values of two masking methods used in MPEG-1 audio coding.
Convention Paper 8051 (Purchase now)
P12-7 Real-Time Speech-Rate Modification Experiments—Adam Kupryjanow, Andrzej Czyzewski, Gdansk University of Technology - Gdansk, Poland
An algorithm designed for real-time speech time scale modification (stretching) is proposed, providing a combination of typical synchronous overlap and add based time scale modification algorithm and signal redundancy detection algorithms that allow to remove parts of the speech signal and replace them with the stretched speech signal fragments. Effectiveness of signal processing algorithms are examined experimentally together with the resulting sound quality.
Convention Paper 8052 (Purchase now)