Sound Classification in Hearing Aids Inspired by Auditory Scene Analysis
- PDF / 1,118,451 Bytes
- 12 Pages / 600 x 792 pts Page_size
- 61 Downloads / 219 Views
ound Classification in Hearing Aids Inspired by Auditory Scene Analysis ¨ Michael Buchler ENT Department, University Hospital Zurich, CH-8091 Zurich, Switzerland Email: [email protected]
Silvia Allegro Phonak AG, CH-8712 Staefa, Switzerland Email: [email protected]
Stefan Launer Phonak AG, CH-8712 Staefa, Switzerland Email: [email protected]
Norbert Dillier ENT Department, University Hospital Zurich, CH-8091 Zurich, Switzerland Email: [email protected] Received 29 April 2004; Revised 16 November 2004 A sound classification system for the automatic recognition of the acoustic environment in a hearing aid is discussed. The system distinguishes the four sound classes “clean speech,” “speech in noise,” “noise,” and “music.” A number of features that are inspired by auditory scene analysis are extracted from the sound signal. These features describe amplitude modulations, spectral profile, harmonicity, amplitude onsets, and rhythm. They are evaluated together with different pattern classifiers. Simple classifiers, such as rule-based and minimum-distance classifiers, are compared with more complex approaches, such as Bayes classifier, neural network, and hidden Markov model. Sounds from a large database are employed for both training and testing of the system. The achieved recognition rates are very high except for the class “speech in noise.” Problems arise in the classification of compressed pop music, strongly reverberated speech, and tonal or fluctuating noises. Keywords and phrases: hearing aids, sound classification, auditory scene analysis.
1.
INTRODUCTION
It was shown in the past that one single setting of the frequency response or of compression parameters in the hearing aid is not satisfying for the user. Kates [1] presented a summary of a number of studies where it was shown that different hearing aid characteristics are desired under different listening conditions. Therefore, modern hearing aids provide typically several hearing programs to account for different acoustic situations, such as quiet environment, noisy environment, music, and so forth. These hearing programs can be activated either by means of a switch at the hearing aid or with a remote control. The manual switching between difThis is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ferent hearing programs is however annoying, as the user has the bothersome task of recognizing the acoustic environment and then switching to the program that best fits this situation. Automatic sensing of the current acoustic situation and automatic switching to the best fitting program would therefore greatly improve the utility of today’s hearing aids. There exist already simple approaches to automatic sound classification in hearing aids, and even though today their performance is not faultless in every listening situation, a field study with one of these approaches has shown that
Data Loading...