Rainer Huber

QUALITY ASSESSMENT OF MULTI-CHANNEL AUDIO PROCESSING SCHEMES BASED ON A BINAURAL AUDITORY MODEL

A perceptual, binaural audio-quality model is introduced. The model was developed for predicting any kinds of perceived spatial quality differences between two audio signals in multi-channel reproduction and audio processing schemes. It employs a recent binaural auditory model as front-end to provide perceptually relevant binaural features for the reference and test audio signal. Correlations between the binaural features of both signals are combined to an overall spatial quality measure by the use of multivariate adaptive regression splines (MARS). Furthermore, a database was generated to train and evaluate the model. The database contains various multi-channel audio signals, which were subjectively assessed in formal listening tests with 15 trained listeners. The results show different model prediction performances depending on the type of quality degradation. Combination of the proposed spatial quality measure with established monaural quality measures improved the predictive power.

Modifications of the MUlti stimulus test with Hidden Reference and Anchor (MUSHRA) for use in audiology

Abstract Objective: Two modifications of the standardised MUlti Stimulus test with Hidden Reference and Anchor (MUSHRA), namely MUSHRA simple and MUSHRA drag&drop, were implemented and evaluated together with the original test method. The modifications were designed to maximise the accessibility of MUSHRA for elderly and technically non-experienced listeners, who constitute the typical target group in hearing aid evaluation. Design: Three MUSHRA variants were assessed based on subjective and objective measures, e.g. test–retest reliability, discrimination ability, time exposure and overall preference. With each method, participants repeated the task to rate the quality of several hearing aid algorithms four times. Study sample: Fifty listeners grouped into five subject classes were tested, including elderly and technically non-experienced participants with normal and impaired hearing. Normal-hearing, technically experienced students served as controls. Results: Both modifications can be used to obtain compatible rating results. Both were preferred over the classical MUSHRA procedure. Technically experienced listeners performed best with the modification MUSHRA drag&drop. Conclusions: The comprehensive comparison of the MUSHRA variants demonstrates that the intuitive modification MUSHRA drag&drop can be generally recommended. However, considering e.g. specific evaluation demands, we suggest a differentiated and careful application of listening test methods.

Comparison of single-microphone noise reduction schemes: can hearing impaired listeners tell the difference?

Abstract Objective: The perceived qualities of nine different single-microphone noise reduction (SMNR) algorithms were to be evaluated and compared in subjective listening tests with normal hearing and hearing impaired (HI) listeners. Design: Speech samples added with traffic noise or with party noise were processed by the SMNR algorithms. Subjects rated the amount of speech distortions, intrusiveness of background noise, listening effort and overall quality, using a simplified MUSHRA (ITU-R, 2003) assessment method. Study sample: 18 normal hearing and 18 moderately HI subjects participated in the study. Results: Significant differences between the rating behaviours of the two subject groups were observed: While normal hearing subjects clearly differentiated between different SMNR algorithms, HI subjects rated all processed signals very similarly. Moreover, HI subjects rated speech distortions of the unprocessed, noisier signals as being more severe than the distortions of the processed signals, in contrast to normal hearing subjects. Conclusions: It seems harder for HI listeners to distinguish between additive noise and speech distortions or/and they might have a different understanding of the term “speech distortion” than normal hearing listeners have. The findings confirm that the evaluation of SMNR schemes for hearing aids should always involve HI listeners.

Evaluation of an automated speech-controlled listening test with spontaneous and read responses

Abstract A method for an automated system for speech audiometry is introduced and evaluated using pre-recorded responses as well as spontaneous utterances produced by listeners during a real measurement. A hearing test is performed under the use of automatic speech recognition (ASR) based on the matrix sentence test, which is used clinically for diagnostics and fitting of hearing devices as well as in psychoacoustic research. The test measures the speech reception threshold (SRT), i.e., the signal-to-noise ratio at which the subject achieves 50% word recognition rate. A major disadvantage of current testing procedures is the requirement of a human expert supervising the test and logging the listener’s responses. An automated system reduces the required resources and therefore provides a tool for frequent assessment of the SRT, which can contribute to an early diagnosis of hearing loss. The accuracy of the ASR-based SRT measurement is compared to results obtained with a human supervisor. To this end, two databases are used that contain either well-controlled read utterances that resemble typical responses during SRT measurements produced by 17 speakers, or spontaneous responses collected during real SRT measurements using ASR. Twenty normal-hearing and seven slightly to moderate hearing-impaired subjects participated in the collection of this spontaneous speech. In order to assess the SRT accuracy for read speech, two simulation schemes are proposed that employ Monte Carlo tests to simulate a listener’s profile and corresponding responses, which are validated with the real measurement data. We show that ASR deletion rates of 0.9% and insertion rates of 2.9% for matrix text words are sufficiently low to obtain accurate SRT measurements in the range of 0.5 dB SNR. This is comparable to the test-retest accuracy obtained by human supervisors. While ASR errors are overestimated when using the controlled speech material in comparison to spontaneous speech, this error type has minimal effect on SRT estimation. Hence, the use of pre-recorded, read speech material is sufficient when evaluating the accuracy of speech-controlled, automated listening tests.

Single-ended prediction of listening effort using deep neural networks

ABSTRACT The effort required to listen to and understand noisy speech is an important factor in the evaluation of noise reduction schemes. This paper introduces a model for Listening Effort prediction from Acoustic Parameters (LEAP). The model is based on methods from automatic speech recognition, specifically on performance measures that quantify the degradation of phoneme posteriorgrams produced by a deep neural net: Noise or artifacts introduced by speech enhancement often result in a temporal smearing of phoneme representations, which is measured by comparison of phoneme vectors. This procedure does not require a priori knowledge about the processed speech, and is therefore single‐ended. The proposed model was evaluated using three datasets of noisy speech signals with listening effort ratings obtained from normal hearing and hearing impaired subjects. The prediction quality was compared to several baseline models such as the ITU‐T standard P.563 for single‐ended speech quality assessment, the American National Standard ANIQUE+ for single‐ended speech quality assessment, and a single‐ended SNR estimator. In all three datasets, the proposed new model achieved clearly better prediction accuracies than the baseline models; correlations with subjective ratings were above 0.9. So far, the model is trained on the specific noise types used in the evaluation. Future work will be concerned with overcoming this limitation by training the model on a variety of different noise types in a multi‐condition way in order to make it generalize to unknown noise types. HighlightsA new single‐ended listening effort prediction method is proposed.The method achieves high correlations (r > 0.9) with subjective ratings.It clearly outperforms standard methods for single‐ended speech quality assessment.

Prediction of perceived sound quality of hearing aids (algorithms) using perceptual models

This contribution presents an overview of basic approaches for predicting the perceived sound quality of hearing aids and hearing aid algorithms using auditory processing models for hearing impaired. Comparison‐based concepts will be considered in particular. The main characteristic of these concepts is the comparison of internal representations (the outputs of the auditory models) of a test and a reference signal. While this approach is straight‐forward and has proven to be successful for the prediction of sound quality of lossy speech and audio processing systems perceived by normal‐hearing listeners, the requirement of a reference representing the optimal quality can pose a problem in the case of hearing aids and hearing‐impaired listeners. The potential and limitations of comparison‐based approaches will be illustrated by example results from different studies obtained with an extended version of the perceptual audio quality model PEMO‐Q.

Perception‐model‐based sound quality prediction for hearing aids: Big effect, small effect, and the optimization of noise reduction algorithms

Objective sound quality models can be used to optimize hearing aid algorithms and their respective parameter settings by numerically predicting the subjective sound quality assessed by humans without having to perform quality measurements with humans concurrently. The big effect prediction considers how the (individual) hearing loss affects the aided or unaided perceived sound quality. The small effect prediction aims at subtle differences at a given amplification level of the hearing aid in order to compare processing parameters. This talk will concentrate on benchmarking different types of objective prediction methods both for big effect and small effect conditions using a wide range of signals and processing conditions, respectively. In addition, the performance of PEMO—Q, an auditory‐model‐based quality prediction scheme developed at Oldenburg University will be discussed, which has been modified to include the effect of hearing impairment and the adaptation to different kinds of test data sets. We wi...