Predicting the impact of hearing aid processing on speech intelligibility

The purpose of this study was twofold: (a) to determine the extent to which 4-channel, slow-acting wide dynamic range amplitude compression (WDRC) can counteract the perceptual effects of reduced auditory dynamic range and (b) to examine the relation between objective measures of speech intelligibility and categorical ratings of speech quality for sentences processed with slow-acting WDRC. Multiband expansion was used to simulate the effects of elevated thresholds and loudness recruitment in normal hearing listeners. While some previous studies have shown that WDRC can improve both speech intelligibility and quality, others have found no benefit. The current experiment shows that moderate amounts of compression can provide a small but significant improvement in speech intelligibility, relative to linear amplification, for simulated-loss listeners with small dynamic ranges (i.e., flat, moderate hearing loss). This benefit was found for speech at conversational levels, both in quiet and in a background of babble. Simulated-loss listeners with large dynamic ranges (i.e., sloping, mild-to-moderate hearing loss) did not show any improvement. Comparison of speech intelligibility scores and subjective ratings of intelligibility showed that listeners with simulated hearing loss could accurately judge the overall intelligibility of speech. However, in all listeners, ratings of pleasantness decreased as the compression ratio increased. These findings suggest that subjective measures of speech quality should be used in conjunction with either objective or subjective measures of speech intelligibility to ensure that participant-selected hearing aid parameters optimize both comfort and intelligibility.

Neural envelope tracking predicts speech intelligibility and hearing aid benefit in children with hearing loss

A battery-operated digital processor connected to microphones and receivers located in left and right ear modules was built and used in a hearing aid field trial. Eight hearing impaired individuals with moderate to moderately severe hearing losses served as subjects. All subjects had symmetric hearing losses and were experienced binaural hearing aid users. Four binaural hearing aid algorithms were programmed into the processor for evaluation in the field trial. The algorithms all equalized the magnitude and phase insertion effects of the ear modules, but differed in their gain prescription. Two prescriptions based on the Articulation Index (AI), one on NAL-R, and a control prescription were evaluated in the two week field trial. Subjects rated each algorithm in seven categories. Objective measures of speech intelligibility in noise, including measures of binaural directional hearing, were taken before and after the field trial. Intelligibility and directional hearing was best with the AI prescriptions, although these prescriptions did not receive the highest subjective ratings. Details of the binaural algorithms and fittings, as well as objective and subjective measures of their benefit will be reported.

Objective. Currently, only behavioral speech understanding tests are available, which require active participation of the person being tested. As this is infeasible for certain populations, an objective measure of speech intelligibility is required. Recently, brain imaging data has been used to establish a relationship between stimulus and brain response. Linear models have been successfully linked to speech intelligibility but require per-subject training. We present a deep-learning-based model incorporating dilated convolutions that operates in a match/mismatch paradigm. The accuracy of the model’s match/mismatch predictions can be used as a proxy for speech intelligibility without subject-specific (re)training. Approach. We evaluated the performance of the model as a function of input segment length, electroencephalography (EEG) frequency band and receptive field size while comparing it to multiple baseline models. Next, we evaluated performance on held-out data and finetuning. Finally, we established a link between the accuracy of our model and the state-of-the-art behavioral MATRIX test. Main results. The dilated convolutional model significantly outperformed the baseline models for every input segment length, for all EEG frequency bands except the delta and theta band, and receptive field sizes between 250 and 500 ms. Additionally, finetuning significantly increased the accuracy on a held-out dataset. Finally, a significant correlation (r = 0.59, p = 0.0154) was found between the speech reception threshold (SRT) estimated using the behavioral MATRIX test and our objective method. Significance. Our method is the first to predict the SRT from EEG for unseen subjects, contributing to objective measures of speech intelligibility.

Speech intelligibility prediction of noisy and processed noisy speech is important in a number of application domains such as hearing instruments and forensics. Most available objective intelligibility measures employ either a signal-to-noise ratio (SNR)-based or correlation-based comparison between frequency bands of the clean and the processed speech. In this paper, we approach the speech intelligibility prediction from the angle of information theory and show that an information theoretic concept provides a unified viewpoint on both the SNR and the correlation based approaches. Two objective intelligibility measures are introduced based on estimated mutual information between the clean speech and the processed speech in the time and the frequency subband domain. Our proposed measures show high correlation with subjective intelligibility measure (i.e. word correct scores) and comparative results with the short-term objective intelligibility measure (STOI).

<p><b>Speech enhancement was, is, and will be the key technology for digital speech transmission. When developing speech enhancement algorithms, the intelligibility or the quality of the processed speech needs to be evaluated. Intelligibility is more fundamental than quality. The evaluation of speech intelligibility can be carried out through subjective listening tests and objective metrics. Carrying out subjective listening tests is time-consuming and costly. Using the objective metrics is more efficient. Thus, there has been an increasing research in speech intelligibility prediction.</b></p> <p> Speech intelligibility can be measured in terms of the information received by a listener. This thesis aims at developing a mutual information-based speech intelligibility predictor (SIP) and using the mutual information-based predictor to assist the fitting process in hearing instruments. To achieve this goal, this thesis carried out three studies. </p> <p>First, it studied the modeling of the transmitted message. For mutual information-based SIPs, the most important thing is to determine the transmitted message. This thesis studied two approaches of modeling of the message: one is the continuous-valued sound, and the other is the discrete-valued linguistic message. Two corresponding SIPs were developed. By comparing their predicted intelligibility results with the psychometric curves, which are the subjective intelligibility scores, it shows that the modeling of discrete-valued message gives a better match to the psychometric curves.</p> <p> Second, based on the modeling of the discrete-valued message, this thesis proposed a mutual information-based SIP. Since the discrete-valued message cannot be obtained from a speech signal, the proposed SIP calculates the mutual information between the clean speech and the received speech, instead of calculating the mutual information between the message and the received speech. The proposed SIP considers frequency correlation for both the clean speech and the received speech. The evaluation results show that the proposed SIP performs better than the existing state-of-the-art mutual information-based SIPs.</p> <p>Third, this thesis proposed an automatic fitting tool for the nonlinear frequency compression (NFC) operator, which is a frequency lowering operator used in hearing instruments. The automatic fitting tool adjusts the parameter in the NFC by maximizing the mutual information between the message and the frequency-lowered speech. To evaluate the automatic fitting tool, the parameter was also searched by listening tests. The results show that the parameter determined by the automatic fitting tool is consistent to the parameter determined by the listening tests.</p>

Eight auditoriums with a proscenium type of stage, which usually operate as dramatic theaters in the city of Sao Paulo, were acoustically surveyed in terms of their adequacy to unassisted speech. Reverberation times, early decay times, and speech levels were measured in different positions, together with objective measures of speech intelligibility. The measurements revealed reverberation time values rather uniform throughout the rooms, whereas significant variations were found in the values of the other acoustical measures with position. The early decay time was found to be better correlated with the objective measures of speech intelligibility than the reverberation time. The results from the objective measurements of speech intelligibility revealed that the speech transmission index STI, and its simplified version RaSTI, are strongly correlated with the early-to-late sound ratio C50 (1 kHz). However, it was found that the criterion value of acceptability of the latter is more easily met than the former. The results from these measurements enable to understand how the characteristics of the architectural design determine the acoustical quality for speech. Measurements of ST1-Gade were made as an attempt to validate it as an objective measure of ‘‘support’’ for the actor. The preliminary diagnosing results with ray tracing simulations will also be presented.

This paper examines the improvement that can be attained with perfect knowledge of the sound source directivity pattern and orientation in beamformer designs in the problem of speech acquisition. Data-independent beamformers are derived through formulation of a constrained optimization problem with a unity-gain constraint. Using computer simulation, these beamforming schemes are compared to the delay and sum (DS) beamformer and the best single sensor in a reverberant room environment. Criteria used to measure performance are: (1) the direct to reverberant ratio, to assess extent of reverberation suppression; and (2) an objective measure of speech intelligibility called the speech transmission index (STI). For human-speaker source directivity, simulation results show that modest improvements to performance are obtainable.

Background: Following the development of new hearing technologies, assessment of speech intelligibility in hearing-impaired children is of great interest. The main purpose of this study was to compare speech intelligibility and auditory perception abilities in children with normal-hearing (NH) and children with hearing aid (HA) and cochlear implant (CI). Methods: This analytic cross-sectional study consisted of 60 Persian-speaking children aged 5 to 7-years. Participants were classified into 3 groups of 20 people, including NH (mean age, 71.70±5.05 months), CI (mean age, 72.60±8.20 months), and HA (mean age, 71.45±10.56 months) children. The speech intelligibility rating (SIR) and categories of auditory performance (CAP) tests were conducted for all children to measure their speech intelligibility and auditory perception, respectively. A one-way analysis of variance (ANOVA) test was used to compare CAP and SIR scores among the 3 groups. Results: The mean SIR score in the NH children was significantly higher than the HA (p=0.002) and CI (p=0.009) groups. However, these differences between the HA and CI children were not significant (p=0.885). We found a significant difference between the CAP scores in the 3 groups (p=0.038). Furthermore, the post hoc analysis results indicated that the mean CAP scores in NH children were significantly higher than the HA and CI participants. Based on the results, the speech intelligibility and auditory performance abilities in NH children were significantly greater than the hearing-impaired (CI and HA) children. However, these abilities between HA and CI users were not significantly different. Conclusion: Based on the results, the early acoustical amplification on auditory and speech functions in children with hearing loss is of paramount importance.

Many objective measures have been reported to predict speech intelligibility in noise, most of which were designed and evaluated with English speech corpora. Given the different perceptual cues used by native listeners of different languages, examining whether there is any language effect when the same objective measure is used to predict speech intelligibility in different languages is of great interest, particularly when non-linear noise-reduction processing is involved. In the present study, an extensive evaluation is taken of objective measures for speech intelligibility prediction of noisy speech processed by noise-reduction algorithms in Chinese, Japanese, and English. Of all the objective measures tested, the short-time objective intelligibility (STOI) measure produced the most accurate results in speech intelligibility prediction for Chinese, while the normalized covariance metric (NCM) and middle-level coherence speech intelligibility index ( CSIIm) incorporating the signal-dependent band-importance functions (BIFs) produced the most accurate results for Japanese and English, respectively. The objective measures that performed best in predicting the effect of non-linear noise-reduction processing in speech intelligibility were found to be the BIF-modified NCM measure for Chinese, the STOI measure for Japanese, and the BIF-modified CSIIm measure for English. Most of the objective measures examined performed differently even under the same conditions for different languages.

When fitting hearing aids, there are three guidelines that should always be followed by the dispenser: restore audibility; so that the amplified sound is above the user's threshold, limit the output; so that the amplified signal does not exceed the user's discomfort level, and do no harm; so that the amplified signal is not unintentionally or undesirably altered by the hearing aid. The first two of these three guidelines are obvious. The first is to present sounds above the user's threshold of hearing in order to create an acoustic environment with the maximum amount of audible speech cues. The second ensures that discomfort level is not exceeded, so that the user does not wince or remove the hearing aids when loud sounds are present. The third guideline—do no harm—is less obvious. This indicates that it is important that the hearing aids produce no alteration of the signal other than that which the fitter intends. This is a subtle way of saying that, among other things, well-fitted hearing aids should not distort the desired sound. The topic of distortion is the subject of this issue of Trends in Amplification. The intention of this issue is to explain why distortion occurs in some hearing aids; to explain how distortion is measured; to summarize what can be done to prevent distortion; and to summarize the perception of distortion by the hearing aid wearer. Distortion in hearing aids can be broadly defined as the generation of undesired audible components that are present in the output, but which are not present in the input. When distortion occurs, hearing aids produce undesired elements at the output through the interaction of the processed signal with some internal non-linear mechanism. These undesired components may interfere to some degree or other with the reception of sound by the listener. If these added elements are small compared to the overall signal level, they may effectively cause no interference at all. If they are large, they can be so disruptive to the listener that the desired sound becomes irritating or even incomprehensible. All audio systems inevitably contain some amount of distortion. The practical problem for the listener is the type of distortion that is present and the level that is acceptable or tolerable in hearing aids before the distortion becomes disruptive to speech intelligibility and sound quality. It is known that highly distorted speech in quiet can remain intelligible (Licklider, 1946); however, as dispensers often experience, intelligibility is not the only measure of whether or not a user will accept and wear hearing aids. Gabrielsson and Sjogren (1979b) have shown that overall perceived sound quality is important to users when selecting a hearing aid. Punch (1978) has shown that listeners with mild to moderate sensorineural hearing losses retain the same ability to make distinctions in sound quality judgments as listeners with normal hearing. This implies that good sound quality is as equally important and desired by listeners with a hearing impairment as it is for listeners with normal hearing. Three characteristics of distortion in hearing aids modify the sound delivered to the user: the type of distortion, the relative amount of distortion at different frequencies, and the variation of distortion with different input levels. Since the most important goal for fitting hearing aids is to restore or facilitate communication ability, undistorted sound is important for optimum speech intelligibility and sound quality. However, it is important to also recognize that there are other reasons to provide undistorted amplification of sound. For example, undistorted and pleasant reproduction of music may be an important sensory experience for some listeners. This issue concentrates on different types of undesired modification to the waveshape of the sound, primarily by total harmonic distortion (THD) and intermodulation distortion (IMD). However, in a more general sense, any undesired component added to the sound by hearing aids is a form of distortion. Thus, this issue also discusses sound generated by hearing aids with no signal present at the microphone. This artifact is more commonly called internal noise. Internal noise fits within the definition of distortion given earlier, since internal noise is an undesired product generated in the output of hearing aids that is not present at the input. In some cases, distortion and internal noise are so closely linked that they cannot always be separated into discrete entities. For example, instability in the frequency and shape of the clock pulses in a Class D output stage (often loosely called clock jitter) results in distortion of the signal. However, this artifact may be audibly revealed in the output sound primarily as noise, rather than as distortion of the waveform. According to this broad definition of distortion, undesired noises such as “motorboating” and other acoustic feedback oscillations can also be considered to be forms of distortion. However, acoustic feedback and other similar audible artifacts occurring within hearing aids will not be discussed here. These types of distortion have been discussed in detail by Agnew (1996b) in a previous issue of Trends in Amplification. In this issue distortion is discussed in two general categories: Undesired modification of the waveform of the incoming sound by some mechanism occurring within the hearing aids. The severity of this effect may or may not vary with the acoustic level of the incoming sound. This type of audible artifact is generally what most dispensers refer to as “distortion”. This will be the subject of Part I of this issue, and will follow the definition used for distortion in this text. The production of undesired audible components in the output sound, regardless of whether or not there is any acoustic input to the hearing aid. This is generally called “noise”. This will be the subject of Part II of this issue. The first part of this issue discusses the causes and effect of distortion. First, different types of distortion are described, followed by an explanation of how they are measured. This is followed by a discussion of the sources of distortion that may be present in hearing aids and the various mechanisms by which distortion is created. The final section of Part I discusses the effects of distortion on the hearing aid wearer. For the reader having difficulty understanding the technical concepts in the first part of this tutorial, a good overview from a different perspective has been presented by Kuk (1996). The second part of this issue discusses the causes and effects of internal noise in hearing aids. This part starts with a discussion of sources of internal noise and continues with a description of methods for the measurement of internal noise. This part concludes with a discussion of the perception of internal noise and its effects on the user of hearing aids. At the end of the text is an intentionally long list of references, which is intended to serve as a reasonably comprehensive resource for the reader seeking further information on distortion and noise in hearing aids.

Considerations on Health Literacy and Hearing Aid Use

In order to help reduce hearing loss, the Occupational Safety and Health Administration regulates noise levels in work environments. However, hearing aids are the primary rehabilitative service provided for individuals with an occupational hearing loss. Very little is being done to monitor hearing aid use in the work environment. Noise which may be safe to an unaided ear can amplified to levels that are damaging to the ear when a hearing aid is being worn. However, it is necessary for some individuals to wear amplification in these noisy environments for safety reasons. As a consequence it is important that these individuals be able to understand speech in the presence of industrial noise while wearing amplification. The purpose of this study was to determine if there is a significant difference in speech intelligibility between linear hearing aids and different types of non-linear hearing aids when they are used in the presence of industrial noise. Twenty-four normal hearing subjects were selected for this study. Each subject was ask to identify words in four CID W-22 lists which had been recorded through a linear hearing aid and two different non-linear hearing aids. Test results showed significantly better word· recognition for the linear in quiet condition over all other conditions. Significantly higher scores were obtained for the TILL condition than were obtained for the Linear in noise and the BILL condition. These preliminary results suggest that an individual wearing amplification in a noisy work environment would benefit with a TILL circuit. The TILL circuit would provide better speech intelligibility in this type of environment. Therefore, providing a safer work environment for the hearing aid user.

This thesis is concerned with noise reduction in hearing aids. Hearing - impaired listeners and hearing - impaired users have great difficulty understanding speech in a noisy background. This problem has motivated the development and the use of noise reduction algorithms to improve the speech intelligibility in hearing aids. In this thesis, two noise reduction algorithms for single channel hearing instruments are presented, evaluated using objective and subjective tests. The first noise reduction algorithm, conventional Spectral Subtraction, is simulated using MATLAB 6.5, R13. The second noise reduction algorithm, Spectral Subtraction in wavelet domanin is introduced as well. This algorithm is implemented off line, and is compared with conventional Spectral Subtraction. A subjective evaluation demonstrates that the second algorithm has additional advantages in speech intelligibility, in poor listening conditions relative to conventional Spectral Subtraction. The subjective testing was performed with normal hearing listeners, at Ryerson University. The objective evaluation shows that the Spectral Subtraction in wavelet domain has improved Signal to Noise Ratio compared to conventional Spectral Subtraction.

Aided cortical response, speech intelligibility, consonant perception and functional performance of young children using conventional amplification or nonlinear frequency compression