Adaptive noise suppression for a dual-microphone hearing aid: Supresión adaptativa del ruido para un auxiliar auditivo con micrófono dual

An adaptive beamformer for behind-the-ear dual-microphone hearing aids has been optimized for speech intelligibility enhancement in the presence of disturbing sounds or noise. The noise reduction approach is based on the scheme presented by Vanden Berghe and Wouters (1998). A real-time implementation of the signal processing is realized in Audallion, a wearable, small digital signal processing (DSP) platform. After physical evaluation, speech-in-noise intelligibility tests have been carried out on three normally-hearing and two hearing-impaired subjects. A significant speech reception threshold improvement of 11.3 dB was obtained in a moderately reverberant environment for one jammer sound source (steady speech-weighted noise or multi-talker babble) in a direction of 90° relative to the direction of the speech.

Sumario

Se perfeccionó el formador de haz adaptable para el micrófono dual de los auxiliares auditivos tipo curveta, con el fin de mejorar la inteligibilidad en ambiente ruidoso. El enfoque para disminuir el ruido está basado en el esquema presentado por Vanden Berghe y Wouters (1998). Se realizó un procesamiento de señal en tiempo real en Audillion, una pequeña plataforma de procesamiento digital de señal (DSP). Después de una evaluación fisica, se llevaron a cabo pruebas de inteligibilidad del lenguaje en ambiente ruidoso en tres sujetos normoyentes y dos hipoacúsicos. Se obtuvo una mejoria significativa en el umbral de recepción del habla de 11.3dB, en un ambiente moderadamente reverberante para una fuente de interferencia (ruido de habla estable o balbuceo multiple) en una dirección de 90° en relación con la dirección del lenguaje     

Evaluation of phoneme compression schemes designed to compensate for temporal and spectral masking in background noise

The effect of phonemic compression has been studied on speech intelligibility in background noise in hearing-impaired listeners with moderate-to-severe high-frequency losses. One configuration, anti-upward-spread-of-masking (anti-USOM) focuses on a release from spectral masking of high-frequency speech cues by selective spectral tilting. Release from temporal masking is the main goal of a second configuration, high-pass filtered compression (HFC), which reduces the speech modulations within a high-pass filtered compression channel. Speech intelligibility was measured with consonant-vowel-consonant (CVC) words in a multi-talker babble and a single-talker background noise. Anti-USOM has a significant negative effect on the phoneme scores in background noise. HFC compression tends to improve vowel intelligibility in a single-talker background noise, especially for the listeners with a relatively poor speech score. In a multi-talker babble noise the effects of HFC compression tend to be negative. It can be concluded that no significant release from spectral or temporal masking is obtained by the applied processing.     

An objective measure for selecting microphone modes in OMNI/DIR hearing aid circuits

OBJECTIVES: Studies have shown that listener preferences for omnidirectional (OMNI) or directional (DIR) processing in hearing aids depend largely on the characteristics of the listening environment, including the relative locations of the listener, signal sources, and noise sources; and whether reverberation is present. Many modern hearing aids incorporate algorithms to switch automatically between microphone modes based on an analysis of the acoustic environment. Little work has been done, however, to evaluate these devices with respect to user preferences, or to compare the outputs of different signal processing algorithms directly to make informed choices between the different microphone modes. This study describes a strategy for automatically switching between DIR and OMNI microphone modes based on a direct comparison between acoustic speech signals processed by DIR and OMNI algorithms in the same listening environment. In addition, data are shown regarding how a decision to choose one microphone mode over another might change as a function of speech to noise ratio (SNR) and spatial orientation of the listener. DESIGN: Speech and noise signals were presented at a variety of SNR's and in different spatial orientations relative to a listener's head. Monaural recordings, made in both OMNI and DIR microphone processing modes, were analyzed using a model of auditory processing that highlights the spectral and temporal dynamics of speech. Differences between OMNI and DIR processing were expressed in terms of a modified spectrotemporal modulation index (mSTMI) developed specifically for this hearing aid application. Differences in mSTMI values were compared with intelligibility measures and user preference judgments made under the same listening conditions. RESULTS: A comparison between the results of the mSTMI analyses and behavioral data (intelligibility and preference judgments) showed excellent agreement, especially in stationary noise backgrounds. In addition, the mSTMI was found to be sensitive to changes in SNR as well as spatial orientation of the listener relative to signal and noise sources. Subsequent mSTMI analyses on hearing aid recordings obtained from real-life environments with more than one talker and modulated noise backgrounds also showed promise for predicting the preferred microphone setting in varied and complex listening environments.     

Intelligibility of modified speech for young listeners with normal and impaired hearing.

Exposure to modified speech has been shown to benefit children with language-learning impairments with respect to their language skills (M. M. Merzenich et al., 1998; P. Tallal et al., 1996). In the study by Tallal and colleagues, the speech modification consisted of both slowing down and amplifying fast, transitional elements of speech. In this study, we examined whether the benefits of modified speech could be extended to provide intelligibility improvements for children with severe-to-profound hearing impairment who wear sensory aids. In addition, the separate effects on intelligibility of slowing down and amplifying speech were evaluated. Two groups of listeners were employed: 8 severe-to-profoundly hearing-impaired children and 5 children with normal hearing. Four speech-processing conditions were tested: (1) natural, unprocessed speech; (2) envelope-amplified speech; (3) slowed speech; and (4) both slowed and envelope-amplified speech. For each condition, three types of speech materials were used: words in sentences, isolated words, and syllable contrasts. To degrade the performance of the normal-hearing children, all testing was completed with a noise background. Results from the hearing-impaired children showed that all varieties of modified speech yielded either equivalent or poorer intelligibility than unprocessed speech. For words in sentences and isolated words, the slowing-down of speech had no effect on intelligibility scores whereas envelope amplification, both alone and combined with slowing-down, yielded significantly lower scores. Intelligibility results from normal-hearing children listening in noise were somewhat similar to those from hearing-impaired children. For isolated words, the slowing-down of speech had no effect on intelligibility whereas envelope amplification degraded intelligibility. For both subject groups, speech processing had no statistically significant effect on syllable discrimination. In summary, without extensive exposure to the speech processing conditions, children with impaired hearing and children with normal hearing listening in noise received no intelligibility advantage from either slowed speech or envelope-amplified speech.     

Factors affecting predicted speech intelligibility with cochlear implants in an auditory model for electrical stimulation

A model of the auditory response to stimulation with cochlear implants (CIs) was used to predict speech intelligibility in electric hearing. The model consists of an auditory nerve cell population that generates delta pulses as action potentials in response to temporal and spatial excitation with a simulated CI signal processing strategy. The auditory nerve cells are modeled with a leaky integrate-and-fire model with membrane noise. Refractory behavior is introduced by raising the threshold potential with an exponentially decreasing function. Furthermore, the action potentials are delayed to account for latency and jitter. The action potentials are further processed by a central model stage, which includes spatial and temporal integration, resulting in an internal representation of the sound presented. Multiplicative noise is included in the internal representations to limit resolution. Internal representations of complete word sets for a sentence intelligibility test were computed and classified using a Dynamic-Time-Warping classifier to quantify information content and to estimate speech intelligibility. The number of auditory nerve cells, the spatial spread of the electrodes' electric field, and the internal noise intensity were found to have a major impact on the modeled speech intelligibility, whereas the influence of refractory behavior, membrane noise, and latency and jitter was minor.     

Performance with an adaptive frequency response hearing aid in a sample of elderly hearing-impaired listeners.

This study assessed the efficacy of an adaptive frequency response hearing aid (AFR) for improving speech perception ability in noise among a group of elderly hearing-impaired listeners. A speech recognition task, self-assessed speech intelligibility task, and qualitative judgment task were administered to examine subtle differences in the effects of the AFR "signal processing" versus linear amplification. Group scores showed statistically significant improvement with AFR processing on the speech recognition task involving high-predictability sentences, but not on any other measures. However, there was a trend toward improved scores with AFR processing for low-predictability sentences as well. These results suggest that AFR circuitry may be most useful for enhancing recognition of speech in high-cue contexts. Wide individual subject variability was observed on all measures. This demonstrates the importance of evaluating the effectiveness of noise reduction hearing aids on an individual basis and with more than one task.     

Effects of Noise Reduction on AM and FM Perception

The goal of noise reduction (NR) algorithms in digital hearing aid devices is to reduce background noise whilst preserving as much of the original signal as possible. These algorithms may increase the signal-to-noise ratio (SNR) in an ideal case, but they generally fail to improve speech intelligibility. However, due to the complex nature of speech, it is difficult to disentangle the numerous low- and high-level effects of NR that may underlie the lack of speech perception benefits. The goal of this study was to better understand why NR algorithms do not improve speech intelligibility by investigating the effects of NR on the ability to discriminate two basic acoustic features, namely amplitude modulation (AM) and frequency modulation (FM) cues, known to be crucial for speech identification in quiet and in noise. Here, discrimination of complex, non-linguistic AM and FM patterns was measured for normal hearing listeners using a same/different task. The stimuli were generated by modulating 1-kHz pure tones by either a two-component AM or FM modulator with patterns changed by manipulating component phases. Modulation rates were centered on 3 Hz. Discrimination of AM and FM patterns was measured in quiet and in the presence of a white noise that had been passed through a gammatone filter centered on 1 kHz. The noise was presented at SNRs ranging from ?6 to +12 dB. Stimuli were left as such or processed via an NR algorithm based on the spectral subtraction method. NR was found to yield small but systematic improvements in discrimination for the AM conditions at favorable SNRs but had little effect, if any, on FM discrimination. A computational model of early auditory processing was developed to quantify the fidelity of AM and FM transmission. The model captured the improvement in discrimination performance for AM stimuli at high SNRs with NR. However, the model also predicted a relatively small detrimental effect of NR for FM stimuli in contrast with the average psychophysical data. Overall, these results suggest that the lack of benefits of NR on speech intelligibility is partly caused by the limited effect of NR on the transmission of narrowband speech modulation cues.     

Effect of Speech Degradation on Top-Down Repair: Phonemic Restoration with Simulations of Cochlear Implants and Combined Electric–Acoustic Stimulation

The brain, using expectations, linguistic knowledge, and context, can perceptually restore inaudible portions of speech. Such top-down repair is thought to enhance speech intelligibility in noisy environments. Hearing-impaired listeners with cochlear implants commonly complain about not understanding speech in noise. We hypothesized that the degradations in the bottom-up speech signals due to the implant signal processing may have a negative effect on the top-down repair mechanisms, which could partially be responsible for this complaint. To test the hypothesis, phonemic restoration of interrupted sentences was measured with young normal-hearing listeners using a noise-band vocoder simulation of implant processing. Decreasing the spectral resolution (by reducing the number of vocoder processing channels from 32 to 4) systematically degraded the speech stimuli. Supporting the hypothesis, the size of the restoration benefit varied as a function of spectral resolution. A significant benefit was observed only at the highest spectral resolution of 32 channels. With eight channels, which resembles the resolution available to most implant users, there was no significant restoration effect. Combined electric-acoustic hearing has been previously shown to provide better intelligibility of speech in adverse listening environments. In a second configuration, combined electric-acoustic hearing was simulated by adding low-pass-filtered acoustic speech to the vocoder processing. There was a slight improvement in phonemic restoration compared to the first configuration; the restoration benefit was observed at spectral resolutions of both 16 and 32 channels. However, the restoration was not observed at lower spectral resolutions (four or eight channels). Overall, the findings imply that the degradations in the bottom-up signals alone (such as occurs in cochlear implants) may reduce the top-down restoration of speech.     

Binaural masking level difference for speech signals in noise

This study investigates the binaural masking level difference in both detection and recognition of speech embedded in noise when the signal is phase-shifted 180 degrees between the left and the right ear. Swedish spondaic words masked by noise werc presented binaurally to 10 normally hearing subjects. The results are presented in terms of speech detection threshold, speech recognition threshold, a psychometric function of speech recognition versus signal-to-noise ratio, and speech intelligibility as a function of binaural intelligibility level difference. Both speech detection threshold and speech intelligibility improved when the speech signal was phase-shifted 180 degrees. The binaural intelligibility level difference ranged from 5.7 to 7.7 dB.     

New cochlear implant coding strategy for tonal language speakers

Accurate pitch perception on the basis of fundamental frequency patterns is essential for the processing of lexical tones in tonal languages such as Cantonese. Speech intelligibility in Cantonese-speaking CI recipients was compared using current signal processing strategies, which typically result in poor pitch perception, and a new strategy, known as the multi-channel envelope modulation (MEM) strategy, was designed to enhance temporal periodicity cues to the fundamental frequency. Performance of nine postlingually hearing-impaired adult cochlear implant users was measured twice using each strategy, initially after a four week trial, and again after two weeks of use with each strategy. Speech intelligibility in speech-spectrum shaped noise was measured using the Cantonese hearing in noise test. A fixed noise level of 65 dB A was used and the signal-to-noise ratios were fixed at either +10, +15, or +20 dB, depending on the baseline performance of individual subjects using the clinical processor. Self-reported benefit in 18 listening situations and overall preference for strategies were obtained at the end of these trial periods. Results showed poorer speech intelligibility with CIS while results obtained using ACE and MEM were comparable. Unfamiliar place coding might have contributed to poorer performance using CIS. Self-reported benefit across strategies did not differ in most listening situations. Participants preferred ACE for listening overall in daily situations, and a few preferred MEM in noise. Whilst the results did not demonstrate any advantages for speech recognition in noise when using MEM compared to ACE, no degradation in performance was observed. This implies that the form of processing employed by MEM retains similar segmental information to that provided by ACE and that potentially, future variations/optimizations of MEM may lead to some improvement in tone perception.     

The Intelligibility of Interrupted Speech: Cochlear Implant Users and Normal Hearing Listeners

Compared with normal-hearing listeners, cochlear implant (CI) users display a loss of intelligibility of speech interrupted by silence or noise, possibly due to reduced ability to integrate and restore speech glimpses across silence or noise intervals. The present study was conducted to establish the extent of the deficit typical CI users have in understanding interrupted high-context sentences as a function of a range of interruption rates (1.5 to 24 Hz) and duty cycles (50 and 75 %). Further, factors such as reduced signal quality of CI signal transmission and advanced age, as well as potentially lower speech intelligibility of CI users even in the lack of interruption manipulation, were explored by presenting young, as well as age-matched, normal-hearing (NH) listeners with full-spectrum and vocoded speech (eight-channel and speech intelligibility baseline performance matched). While the actual CI users had more difficulties in understanding interrupted speech and taking advantage of faster interruption rates and increased duty cycle than the eight-channel noise-band vocoded listeners, their performance was similar to the matched noise-band vocoded listeners. These results suggest that while loss of spectro-temporal resolution indeed plays an important role in reduced intelligibility of interrupted speech, these factors alone cannot entirely explain the deficit. Other factors associated with real CIs, such as aging or failure in transmission of essential speech cues, seem to additionally contribute to poor intelligibility of interrupted speech.     

Using response time to speech as a measure for listening effort

Abstract

Objective: Speech signals that do not differ in intelligibility might differ in listening effort. This study aimed to investigate the effect of background noise on response time to intelligible speech. Design: We added various amounts of stationary noise to spoken digit triplets and measured the influence of noise on the response time for both an identification and an arithmetic task: Task 1 ‘identify the final digit in a triplet’, and Task 2 ‘calculate the sum of the initial and the final digits in a triplet.’ Study sample: Twelve normal-hearing participants with a mean age of 30.6 years (range: 28–44 years). Results: Response time increased with lower (i.e. worse) signal to noise ratios for both tasks, even for signal to noise ratios with almost maximum intelligibility (close to 100%). The response time during the arithmetic task was more affected by the noise than during the identification task, but the arithmetic task demonstrated higher variance. Conclusions: The response time to digit triplets reduces significantly for increasing signal to noise ratios, even where speech intelligibility is optimal. These differences in response time might be related to listening effort and as such might be used to evaluate hearing-aid signal processing at positive SNRs.     

An examination of speech recognition in a modulated background and of forward masking in younger and older listeners.

PURPOSE: To compare speech intelligibility in the presence of a 10-Hz square-wave noise masker in younger and older listeners and to relate performance to recovery from forward masking. METHOD: The signal-to-noise ratio required to achieve 50% sentence identification in the presence of a 10-Hz square-wave noise masker was obtained for each of the 8 younger/older listener pairs. Listeners were matched according to their quiet thresholds for frequencies from 600 to 4800 Hz in octave steps. Forward masking was also measured in 2 younger/older threshold-matched groups for signal delays of 2-40 ms. RESULTS: Older listeners typically required a significantly higher signal-to-noise ratio than younger listeners to achieve 50% correct sentence recognition. This effect may be understood in terms of increased forward-masked thresholds throughout the range of signal delays corresponding to the silent intervals in the modulated noise (e.g., <50 ms). CONCLUSIONS: Significant differences were observed between older and younger listeners on measures of both speech intelligibility in a modulated background and forward masking over a range of signal delays (0-40 ms). Age-related susceptibility to forward masking at relatively short delays may reflect a deficit in processing at a fairly central level (e.g., broader temporal windows or less efficient processing).     

Comparison of performance with wide dynamic range compression and linear amplification

This study compared subject performance and preference using a compression-limiting hearing aid set to linear amplification (program 1) and wide dynamic range compression (WDRC, program 2). The frequency responses of the hearing aid were matched to a 65 dB SPL signal and maximum output to a 90 dB SPL signal. Twenty subjects with moderate to moderately severe sensorineural hearing loss were tested. Speech recognition scores and speech reception thresholds were obtained both in quiet and in noise. Subjective preference for WDRC or linear amplification was measured via a paired-comparison procedure on "loudness appropriateness," "clarity," and "pleasantness" to continuous discourse presented in quiet and in noise. Results suggested that WDRC yielded better speech intelligibility in quiet for low-level signals and no difference in speech intelligibility in noise compared to linear amplification. Subjects preferred WDRC for loudness to both high- and low-level signals and for pleasantness to high-level signals.     

Effects of noise and filtering on the intelligibility of speech produced during simultaneous communication

This study investigated the effects of noise and filtering on the intelligibility of speech produced during simultaneous communication (SC). Four normal hearing, experienced sign language users were recorded under SC and speech alone (SA) conditions speaking Boothroyd’s forced-choice phonetic contrast material designed for measurement of speech intelligibility. Twenty-four normal hearing listeners audited the speech samples produced by the four speakers under the SC and SA conditions, three listeners in noise and three listeners in filtered listening conditions for each of the four speakers. Although results indicated longer sentence durations for SC than SA, the data showed no difference in the intelligibility of speech produced during SC versus speech produced during SA under either the noise or filtered listening condition, nor any difference in pattern of phonetic contrast recognition errors between the SA and SC speech samples in either listening condition. This conclusion is consistent with previous research indicating that temporal alterations produced by SC do not produce degradation of temporal or spectral cues to speech intelligibility or disruption of the perception of specific English phoneme segments.

Learning outcomes

As a result of this activity, the participant will be able to (1) describe simultaneous communication; (2) explain the role of simultaneous communication in communication with children who are deaf; (3) discuss methods of measuring speech intelligibility under filtered and noise conditions; and (4) specify the ability of listeners to perceive speech produced during simultaneous communication under noise and filtered listening conditions.     

A comparison of two-channel and single-channel compression hearing aids

Eight subjects with bilateral sensorineural hearing losses took part in a trial comparing listening unaided with listening binaurally through two types of hearing aid, aid A and aid B. Both aids incorporated slow-acting automatic gain control (AGC) operating on the whole speech signal. However, aid A also incorporated two-channel syllabic compression. The two aids were chosen to be as similar as possible in other respects, and both were worn behind the ear. Subjects were tested in a counter-balanced order, and had at least 2 weeks of everyday experience with each aid before testing took place. Performance was evaluated in three ways: by measuring speech intelligibility in quiet for sentences at three peak sound levels, 55, 70 and 85 dB SPL; by measuring the level of speech required for 50% intelligibility (called the SRT) of sentences in two levels of speech-shaped noise, 60 and 75 dB SPL; and by administering questionnaires about experience with the aids in everyday life. Both aid A and aid B improved the intelligibility of speech in quiet relative to unaided listening, particularly at the lowest sound level. However, aid A gave lower (i.e., superior) SRTs in speech-shaped noise than aid B or unaided listening. The questionnaires also indicated that aid A gave better performance in noisy situations. The results strongly suggest that two-channel syllabic compression, combined with slow-acting AGC operating on the whole speech signal, can give superior results to slow-acting AGC alone, particularly in noisy situations.     

Evaluation of model-based versus non-parametric monaural noise-reduction approaches for hearing aids

Abstract

Objective: Single channel noise reduction has been well investigated and seems to have reached its limits in terms of speech intelligibility improvement, however, the quality of such schemes can still be advanced. This study tests to what extent novel model-based processing schemes might improve performance in particular for non-stationary noise conditions. Design: Two prototype model-based algorithms, a speech-model-based, and a auditory-model-based algorithm were compared to a state-of-the-art non-parametric minimum statistics algorithm. A speech intelligibility test, preference rating, and listening effort scaling were performed. Additionally, three objective quality measures for the signal, background, and overall distortions were applied. For a better comparison of all algorithms, particular attention was given to the usage of the similar Wiener-based gain rule. Study sample: The perceptual investigation was performed with fourteen hearing-impaired subjects. Results: The results revealed that the non-parametric algorithm and the auditory model-based algorithm did not affect speech intelligibility, whereas the speech-model-based algorithm slightly decreased intelligibility. In terms of subjective quality, both model-based algorithms perform better than the unprocessed condition and the reference in particular for highly non-stationary noise environments. Conclusion: Data support the hypothesis that model-based algorithms are promising for improving performance in non-stationary noise conditions.     

Callsign Acquisition Test (CAT): speech intelligibility in noise

OBJECTIVE: The study was designed to assess the effects of noise on the intelligibility of speech elements used in the Callsign Acquisition Test (CAT), developed by the U.S. Army Research Laboratory. The CAT consists of 126 test items, or callsigns, each of which is made up of a two-syllable word selected from the 18-item military alphabet (Alpha-Zulu) followed by a one-syllable number (all numbers from 1 to 8, excluding 7). DESIGN: The CAT items were mixed with one of three different types of background noises (pink noise, white noise, and multitalker babble) and presented to 18 listeners. Speech-to-noise ratio for all three noises and the overall level of pink noise were varied in two separate experiments to determine how these variables affected speech intelligibility of the CAT items pronounced by a male talker. CONCLUSIONS: Test results demonstrate speech-to-noise ratio has a significant effect on speech intelligibility of the CAT items under all conditions. Pink noise generated the lowest speech intelligibility scores followed by multitalker babble and then white noise. A change in the overall level of pink noise had only small effect on CAT intelligibility.     

Effects of single-channel phonemic compression schemes on the understanding of speech by hearing-impaired listeners

The effect of digital processing on speech intelligibility was studied in hearing-impaired listeners with moderate to severe high-frequency losses. The amount of smoothed phonemic compression in a high-frequency channel was varied using wide-band control. Two alternative systems were tested to compensate for upward spread of masking (USOM) and to reduce modulations in the high-frequency channel effectively. Consonant-vowel-consonant tests were conducted in a group of 14 subjects using eight different speech-processing settings. Speech intelligibility improved significantly with compression, mainly due to positive effects on the initial-consonant score. Surprisingly, listeners with a smaller residual dynamic range tended to profit less from compression. Compensation for USOM gave an additional improvement of vowel intelligibility. In background noise, consistently negative effects of speech processing were found. The combined use of phonemic compression and USOM compensation is promising in conditions without background noise.     

Is complex signal processing for bone conduction hearing aids useful?

Abstract

Objectives

To establish whether complex signal processing is beneficial for users of bone anchored hearing aids.

Methods

Review and analysis of two studies from our own group, each comparing a speech processor with basic digital signal processing (either Baha Divino or Baha Intenso) and a processor with complex digital signal processing (either Baha BP100 or Baha BP110 power). The main differences between basic and complex signal processing are the number of audiologist accessible frequency channels and the availability and complexity of the directional multi-microphone noise reduction and loudness compression systems.

Results

Both studies show a small, statistically non-significant improvement of speech understanding in quiet with the complex digital signal processing. The average improvement for speech in noise is +0.9 dB, if speech and noise are emitted both from the front of the listener. If noise is emitted from the rear and speech from the front of the listener, the advantage of the devices with complex digital signal processing as opposed to those with basic signal processing increases, on average, to +3.2 dB (range +2.3 … +5.1 dB, p ≤ 0.0032).

Discussion

Complex digital signal processing does indeed improve speech understanding, especially in noise coming from the rear. This finding has been supported by another study, which has been published recently by a different research group.

Conclusions

When compared to basic digital signal processing, complex digital signal processing can increase speech understanding of users of bone anchored hearing aids. The benefit is most significant for speech understanding in noise.