Increase Speech Intelligibility Research Articles

Increasing speech intelligibility in monaural hearing by adding noise at the other ear

Ambient environment noise can affect speech intelligibility in phone communication. This paper investigates the feasibility of increasing speech intelligibility in monaural hearing by adding noise at the other ear. The testing materials were generated by mixing the speech from the English Coordinate Response Measure corpus with three types of environmental noise, where 4 signals to noise ratios in the speech ear and 14 noise levels in the contralateral ear were included. The experimental results show that a proper level of contralateral noise can improve the speech intelligibility when the signal to noise ratio in the speech ear is lower than a certain level, but a large contralateral noise level has the opposite effect. A preliminary explanation for the phenomena is attempted by using a binaural loudness model and some psychoacoustic and physiological facts.

A deep learning based segregation algorithm to increase speech intelligibility for hearing-impaired listeners in reverberant-noisy conditions.

Recently, deep learning based speech segregation has been shown to improve human speech intelligibility in noisy environments. However, one important factor not yet considered is room reverberation, which characterizes typical daily environments. The combination of reverberation and background noise can severely degrade speech intelligibility for hearing-impaired (HI) listeners. In the current study, a deep learning based time-frequency masking algorithm was proposed to address both room reverberation and background noise. Specifically, a deep neural network was trained to estimate the ideal ratio mask, where anechoic-clean speech was considered as the desired signal. Intelligibility testing was conducted under reverberant-noisy conditions with reverberation time T 60 = 0.6 s, plus speech-shaped noise or babble noise at various signal-to-noise ratios. The experiments demonstrated that substantial speech intelligibility improvements were obtained for HI listeners. The algorithm was also somewhat beneficial for normal-hearing (NH) listeners. In addition, sentence intelligibility scores for HI listeners with algorithm processing approached or matched those of young-adult NH listeners without processing. The current study represents a step toward deploying deep learning algorithms to help the speech understanding of HI listeners in everyday conditions.

Automated Subjective Assessment of Speech Intelligibility in Various Listening Modes

In this paper, the results of automated subjective assessment of Ukrainian speech intelligibility are presented. Speech monosyllables of the consonant-vowel-consonant (CVC) type were listened in two modes: through headphones and through acoustic monitors. The assessment was carried out with the help of specially developed software that allowed automating of articulation tests. Speech listening was done for four situations: pure language; speech distorted by noise; speech distorted by reverberation; speech distorted by the combined effect of noise and reverberation. In the first case, speech monosyllables of 3 articulation tables were listened, each of which contained 50 monosyllables. In the second case, speech distorted by the additive noise with the signal-to-noise ratios (SNR) varied in the range ‑15…+10 dB was listened. In this case, models of white, pink and brown noises were used, the masking properties of which are rather well-studied. In the third case, the reverberant speech for reverberation times in the range 0.3…2.7 s was modeled by convolution of pure speech signals with room impulse responces (RIRs) of various rooms, and in the fourth case the joint action of pink noise and reverberation was considered. It turned out that the masking ability of white noise exceeds one for brown noise for SNR less than minus 5 dB, which is not entirely consistent with preliminary predictive estimates. In addition, it turned out that listening to speech distorted by noise through acoustic monitors could lead to a significant increase in the speech intelligibility, compared to the case of listening through headphones. The analysis of possible causes of abnormal increase in speech intelligibility has been carried out. Early reflections, presence of two loudspeakers, binaural listening, psychophysical features of listeners, as well as peculiarities of software and articulatory testing organization were considered as possible reasons of the phenomenon. After correction of the software and some features of articulation tests it turned out that the results of the speech intelligibility estimation almost coincide when listening to the signals through the headphones and through acoustic monitors, if the distance between the listener and acoustic monitors does not exceed 0.6-0.8 meters. At the same time, these corrections did not differ in the behavior of the dependencies of speech intelligibility on the SNR for small (less minus 5 dB) SNR values The general conclusion may be that listening to speech signals distorted by noise and reverberation interferences, performed with the application of the proposed automated system of articulation tests, indicates the performance and high quality of the developed system.Ref. 13, fig. 7.

CODING AND ANALYSIS OF SPEECH IN COCHLEAR IMPLANT: A REVIEW

About 5% of the total population of world has hearing impairment and this is 6.3% in India.The hearing impairment in patients can be corrected by using a hearing aid or cochlear implant (CI) .Hearing aid is a traditional way while CI, which is surgically implanted, is a modern way to correct the hearing impairment. In comparison to hearing aids CI increases speech intelligibility at a greater level. Also speech intelligibility decreases in noise though acceptable in noise-free environments. With the advancement of technology over past decades, it shows increase in speech intelligibility with new emerging speech coding strategies. In this paper the basic speech processing strategies for speech coding and different techniques and algorithms, which aimed to increase the speech intelligibility, are being reviewed. Speech will be analyzed based on some measures like Analysis of Variation (ANOVA), Root Mean Square Error (RMSE) etc. The focus will be on speech perception and intelligibility enhancement.

Subjective Evaluation of the Speech Intelligibility on the Background of Noise and Reverberation

In this paper, the results of speech intelligibility subjective assessment of Ukrainian speech monosyllabic sound combinations against background noise and reverberation through articulation tests are presented. The evaluation was carried out with the help of specially developed software that allowed automating and thus greatly facilitating and accelerating the procedure of articulation tests. Special text and sound tables of the Ukrainian speech monosyllables of the consonant-vowel-consonant (CVC) type were developed for the tests. The recordings of pure signals were made in the muffled room of the Acoustics and Acoustoelectronics Department of the National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute". The recording was performed at a frequency of 22050 Hz and a bit depth of 16 bits. Speech monosyllables were read using a verbal environment. Listening was done for four situations: pure language; speech distorted by noise; speech distorted by reverberation; speech distorted by the combined effect of noise and reverberation. In the first case, speech monosyllables of 3 articulation tables were listened, each of which contained 50 monosyllables. In the second case, speech distorted by the additive noise with the signal-to-noise ratios minus 10 dB, 0 dB and plus 10 dB was listened. In this case, models of white, pink and brown noises were used, the masking property of which is rather well-studied. In the third case, the reverberant speech for reverberation times from 0.3 to 2.7 s was modeled by convolution of pure speech signals with impulse characteristics of various rooms, and in the fourth case the joint action of pink noise and reverberation was considered. It turned out that the masking ability of white noise exceeds one for brown noise for signal-to-noise ratios (SNR) less than minus 5 dB, which is not entirely consistent with previous preliminary predictive estimates. Signals listening had been made by two ways, namely through the headphones and by means of acoustic monitors. The SNR varied in the range ‑10…+10 dB, and the reverberation time varied in the range 0.3…2.7 s. In addition, it turned out that listening to speech distorted by noise through acoustic monitors could lead to a significant increase in the intelligibility of speech, compared to the case of listening through headphones. With a signal-to-noise ratio of minus 10 dB, the values of speech intelligibility scores increased from 0.1-0.3 to 0.85-0.93. Similar results were obtained for the reverberation action: the speech intelligibility increased from 0.65 to 0.94 for the reverberation time of 2.7 s. This speech intelligibility growth can be explained, firstly, by the action of early reflections. Secondly, usage of two loudspeakers as sound sources and also binaural listening can be also considered as reasons of speech intelligibility increase. Nevertheless, these reasons for increasing the intelligibility of speech are hardly the only ones. Therefore, it is necessary to further study the causes of the revealed significant increase in speech intelligibility. Among such reasons can be both features of the developed automated system of articulation tests, and features of the psychophysical state of listeners.Ref. 13, fig. 5.

The optimal noise-rejection threshold for normal and impaired hearing

Binary masking represents a powerful tool for increasing speech intelligibility in noise. An essential aspect involves the local criterion (LC), which defines the signal-to-noise ratio below which time-frequency units are discarded. But binary masking is a victim of its own success in one regard—it produces ceiling sentence intelligibility across a broad range of LC values, making the exact optimal LC value difficult to determine. Further, the optimal value for hearing-impaired (HI) listeners is largely unknown. In the current study, the optimal LC was determined in normal-hearing (NH) and HI listeners using speech materials less likely to produce ceiling effects. The CID W22 words were mixed with noise consisting of recordings from a busy hospital cafeteria, then subjected to ideal binary masking. LC values ranged from -20 to + 5 dB relative to the overall SNR of -8 dB. NH subjects were tested at 65 dBA and HI subjects were tested at 65 dBA plus NAL-RP hearing-aid gains. Preliminary results suggest that the optimal LC is similar for NH and HI listeners. Additional conditions involving different speech materials and noise types suggest that the optimal LC can vary as a function of speech and/or noise type. [Work supported by NIH.]

Speech intelligibility after repair of cleft lip and palate.

Background: Intelligibility refers to understandability of speech; and lack of it can negatively affect children’s overall communication effectiveness. Children with repaired cleft lip and/or cleft palate (CL/P) may experience poor speech intelligibility. This study aimed at evaluating speech intelligibility in children with repaired CL/P who had not been referred to speech-language pathology clinics for early intervention. Methods: Sixty-four monolingual Persian-speaking children, 32 children with repaired CL/P, and 32 controls aged 3 to 5 years participated in this survey. Their speech intelligibility was evaluated through the Persian Speech Intelligibility Test and was normalized on children 3 to 5 years. Each speech sample was heard by a speech and language pathologist (SLP), as well as 2 nonprofessional listeners. Two objective measures of speech intelligibility including Percentage of Consonants Correct (PCC) and Percentage of Intelligible Words (PIW) were used in this research. Results: Children with CL/P were significantly outperformed by their peers in PCC (p= 0.0001) and PIW (p= 0.0001). More than half of the case group had compensatory errors and 40.6% had obligatory errors. The PCC and the PIW were statistically different in children with different rates of hyper nasality (p= 0.001). Conclusion: Speech intelligibility of children with CL/P is impaired due to their articulation disorders (obligatory and compensatory errors). This survey documents the necessity for speech therapy for increasing speech intelligibility in this population.

A Study on the Relationship between the Intelligibility and Quality of Algorithmically-Modified Speech for Normal Hearing Listeners

This study investigates the relationship between the intelligibility and quality of modified speech in noise and in quiet. Speech signals were processed by seven algorithms designed to increase speech intelligibility in noise without altering speech intensity. In three noise maskers, including both stationary and fluctuating noise at two signal-to-noise ratios (SNR), listeners identified keywords from unmodified or modified sentences. The intelligibility performance of each type of speech was measured as the listeners’ word recognition rate in each condition, while the quality was rated as a mean opinion score. In quiet, only the perceptual quality of each type of speech was assessed. The results suggest that when listening in noise, modification performance on improving intelligibility is more important than its potential negative impact on speech quality. However, when listening in quiet or at SNRs in which intelligibility is no longer an issue to listeners, the impact to speech quality due to modification becomes a concern.

Cochlear Implantation in Patients With Usher Syndrome Type IIa Increases Performance and Quality of Life.

Usher syndrome type IIa (USH2a) is characterized by congenital moderate to severe hearing impairment and retinitis pigmentosa. Hearing rehabilitation starts in early childhood with the application of hearing aids. In some patients with USH2a, severe progression of hearing impairment leads to insufficient speech intelligibility with hearing aids and issues with adequate communication and safety. Cochlear implantation (CI) is the next step in rehabilitation of such patients. This study evaluates the performance and benefit of CI in patients with USH2a. Retrospective case-control study to evaluate the performance and benefit of CI in 16 postlingually deaf adults (eight patients with USH2a and eight matched controls). Performance and benefit were evaluated by a speech intelligibility test and three quality-of-life questionnaires. Patients with USH2a with a mean age of 59 years at implantation exhibited good performance after CI. The phoneme scores improved significantly from 41 to 87% in patients with USH2a (p = 0.02) and from 30 to 86% in the control group (p = 0.001). The results of the questionnaire survey demonstrated a clear benefit from CI. There were no differences in performance or benefit between patients with USH2a and control patients before and after CI. CI increases speech intelligibility and improves quality of life in patients with USH2a.

Effect of word familiarity on word intelligibility of four continuous words under long-path echo conditions

The massive earthquake that occurred on 11 March 2011 in Japan demonstrated that the intelligibility of speech presented over mass notification sound systems is often significantly degraded by long-path echoes. This study examines the effect of word familiarity on speech intelligibility in the presence of long-path echoes, in order to increase speech intelligibility in such systems. We performed two experiments using sets of four sequentially connected words (quadruplets), in place of an actual sentence. In Experiment 1, we investigated word intelligibility in the presence of simulated long-path echoes for quadruplets consisting of words with the same word familiarity rank. The results indicated that the intelligibility of high-familiarity words is higher than that of low-familiarity words, irrespective of the number of simulated long-path echoes. In Experiment 2, quadruplets with mixed word familiarity were used to investigate intelligibility under more realistic conditions. The results of Experiment 2 demonstrate that the intelligibility of high-familiarity words is higher than that of low-familiarity words under long-path echo conditions, even when high- and low-familiarity words coexist in one quadruplet. These facts show that high-familiarity words are more robust against the influence of long-path echoes than low-familiarity words, strongly suggesting that announcements presented from mass notification sound systems should consist of high familiarity words as much as possible.

APPLICATION OF THE ANALYTIC HIERARCHY PROCESS (AHP) FOR ACOUSTIC ADAPTATION OF CLASSROOMS

Abstract In the paper we put forward a proposition of acoustic adaptation of classrooms based on the analytical hierarchy process (AHP). The main objective of such an adaptation is to increase speech intelligibility aiming to enhance the teaching quality level. The model is based on four main criteria which have impact both on the teaching process and on the adaptation costs. The said criteria comprise reverberation time, speech transmission index STI, financial cost of the adaptation and non-economic costs. The priorities and significance of the main criteria are identified by way of measureable calculations involving reverberation time, STI, costs and expertise, intangible, non-economic cost estimations. The AHP analysis should be viewed as a framework facilitating decision making processes applied in scientific or engineering analyses. The paper demonstrates the acoustic adaptation method of classrooms in terms of acoustic parameters and implementation costs of such an adaptation. The AHP analysis can help architects and classroom users adopt a more rational approach to further optimization and work out a definite building process which could provide a platform to compare different alternatives on a common and comprehensive basis.

A clinical assessment of cochlear implant recipient performance: implications for individualized map settings in specific environments.

Individual speech intelligibility was measured in quiet and noise for cochlear Implant recipients upgrading from the Freedom to the CP900 series sound processor. The postlingually deafened participants (n = 23) used either Nucleus CI24RE or CI512 cochlear implant, and currently wore a Freedom sound processor. A significant group mean improvement in speech intelligibility was found in quiet (Freiburg monosyllabic words at 50 dBSPL) and in noise (adaptive Oldenburger sentences in noise) for the two CP900 series SmartSound programs compared to the Freedom program. Further analysis was carried out on individual’s speech intelligibility outcomes in quiet and in noise. Results showed a significant improvement or decrement for some recipients when upgrading to the new programs. To further increase speech intelligibility outcomes when upgrading, an enhanced upgrade procedure is proposed that includes additional testing with different signal-processing schemes. Implications of this research are that future automated scene analysis and switching technologies could provide additional performance improvements by introducing individualized scene-dependent settings.

Large-scale training to increase speech intelligibility for hearing-impaired listeners in novel noises.

Supervised speech segregation has been recently shown to improve human speech intelligibility in noise, when trained and tested on similar noises. However, a major challenge involves the ability to generalize to entirely novel noises. Such generalization would enable hearing aid and cochlear implant users to improve speech intelligibility in unknown noisy environments. This challenge is addressed in the current study through large-scale training. Specifically, a deep neural network (DNN) was trained on 10 000 noises to estimate the ideal ratio mask, and then employed to separate sentences from completely new noises (cafeteria and babble) at several signal-to-noise ratios (SNRs). Although the DNN was trained at the fixed SNR of - 2 dB, testing using hearing-impaired listeners demonstrated that speech intelligibility increased substantially following speech segregation using the novel noises and unmatched SNR conditions of 0 dB and 5 dB. Sentence intelligibility benefit was also observed for normal-hearing listeners in most noisy conditions. The results indicate that DNN-based supervised speech segregation with large-scale training is a very promising approach for generalization to new acoustic environments.

An algorithm to increase speech intelligibility for hearing-impaired listeners in entirely novel noises

Supervised speech segregation has been recently shown to improve human speech intelligibility in noise, when trained and tested on similar noises. However, a major challenge involves the ability to generalize to entirely novel noises. Such generalization would enable hearing aid andcochlear implant users to improve speech intelligibility in unknown noisy environments. This challenge is addressed in the current study through large-scale training. Specifically, a deep neural network (DNN) was trained on 10,000 noises to estimate the ideal ratio mask, and then employed to separate sentences from completely new noises (cafeteria and babble) at several signal-to-noise ratios (SNRs). Although the DNN wastrained at the fixed SNR of 2 dB, testing using hearing-impaired listeners demonstrated that speech intelligibility increased substantially following speech segregation using the novel noises and unmatched SNR conditions of 0 dB and 5 dB. Sentence-intelligibility benefit was also observed for normal-hearing listeners in most noisy conditions. The results indicate that DNN-based supervised speech segregation with large-scale training is a very promising approach for generalization to new acoustic environments.

Speech-in-noise enhancement using amplification and dynamic range compression controlled by the speech intelligibility index.

In many speech communication applications, such as public address systems, speech is degraded by additive noise, leading to reduced speech intelligibility. In this paper a pre-processing algorithm is proposed that is capable of increasing speech intelligibility under an equal-power constraint. The proposed AdaptDRC algorithm comprises two time- and frequency-dependent stages, i.e., an amplification stage and a dynamic range compression stage that are both dependent on the Speech Intelligibility Index (SII). Experiments using two objective measures, namely, the extended SII and the short-time objective intelligibility measure (STOI), and a formal listening test were conducted to compare the AdaptDRC algorithm with a modified version of a recently proposed algorithm in three different noise conditions (stationary car noise and speech-shaped noise and non-stationary cafeteria noise). While the objective measures indicate a similar performance for both algorithms, results from the formal listening test indicate that for the two stationary noises both algorithms lead to statistically significant improvements in speech intelligibility and for the non-stationary cafeteria noise only the proposed AdaptDRC algorithm leads to statistically significant improvements. A comparison of both objective measures and results from the listening test shows high correlations, although, in general, the performance of both algorithms is overestimated.

An algorithm to increase speech intelligibility for hearing-impaired listeners in novel segments of the same noise type.

Machine learning algorithms to segregate speech from background noise hold considerable promise for alleviating limitations associated with hearing impairment. One of the most important considerations for implementing these algorithms into devices such as hearing aids and cochlear implants involves their ability to generalize to conditions not employed during the training stage. A major challenge involves the generalization to novel noise segments. In the current study, sentences were segregated from multi-talker babble and from cafeteria noise using an algorithm that employs deep neural networks to estimate the ideal ratio mask. Importantly, the algorithm was trained on segments of noise and tested using entirely novel segments of the same nonstationary noise type. Substantial sentence-intelligibility benefit was observed for hearing-impaired listeners in both noise types, despite the use of unseen noise segments during the test stage. Interestingly, normal-hearing listeners displayed benefit in babble but not in cafeteria noise. This result highlights the importance of evaluating these algorithms not only in human subjects, but in members of the actual target population.

Reducing cognitive demands on listeners by speaking clearly in noisy places

Listeners have more difficulty identifying spoken words in noisy environments when those words have many phonological neighbors (i.e., similar-sounding words in the lexicon) than when they have few phonological neighbors. This difficulty appears to be exacerbated in old age, where reductions in inhibitory control presumably make it more difficult to cope with competition from similar-sounding words. Fortunately, word recognition in noise can generally be improved for a wide range of listeners (e.g., younger and older adults, individuals with and without hearing impairment) when speakers adopt a clear speaking style. This study investigated whether clear speech, in addition to generally increasing speech intelligibility, also reduces the inhibitory demands associated with identifying lexically difficult words in noise for younger and older adults. The results show that, indeed, the difference between rates of identification for words with many versus few neighbors was eliminated when those words were produced in clear speech. Data on the roles of individual differences (e.g., hearing, working memory, and inhibitory control) that may contribute to word identification in noise will also be presented.

My Client Talks! Do I Still Need to Consider AAC in my Treatment Planning? Speech Supplementation Strategies: AAC for Clients Who Talk!

People with motor speech disorders often persist in using their natural speech to communicate despite low intelligibility and frequent communication breakdowns. Their speech may be functional in some situations, such as with familiar partners in quiet settings, but not across the variety of communication situations they face. Speech supplementation strategies, such as alphabet supplementation, allow speakers to augment their speech with additional cues. For example, alphabet supplementation, which means the speaker points to the first letter of each word while speaking, has been shown to increase speech intelligibility at the word level with an average gain of 11% and at the sentence level with an average gain of 26% (Hanson, Yorkston, & Beukelman, 2004). This form of AAC can help speakers bridge the gap between ineffective communication with natural speech and total reliance on alternative communication methods (Hanson, Yorkston, & Britton, 2011). The purpose of this article is to summarize the evidence in the area of speech supplementation, provide direction to integrate the strategies into treatment planning, and to suggest research directions for this rich area of study.

Speech intelligibility can improve rapidly during exposure to a novel acoustic environment

In natural listening environments, the background noises and the acoustic spaces they occupy can vary greatly, both in their characteristics and in their impact on speech intelligibility. It has been suggested that listeners build up a statistical representation of ongoing noises; however, listeners can move around a room or move from one room to another, constantly changing both the background noises and the reverberation around them. If listeners are to make use of such statistics to aid in speech intelligibility, they must engage in a constant adaptation. We asked normal and hearing-impaired listeners to identify in real-time a stream of random target words in a contiguous series of novel acoustic environments lasting 13 s each, thus measuring the extent and time-course of changes in speech intelligibility that listeners experience in a new environment. The results from our task demonstrate that most listeners experience a rapid increase in speech intelligibility over several seconds of exposure to certain acoustic environments, but not to others. This suggests that there are classes or types of noises and reverberations that can be adapted to over a short time course and others that cannot. [Work supported by the MRC and the CSO.]

Does acoustic fundamental frequency information enhance cochlear implant performance?

ObjectivesLow-frequency information via an acoustic aid has been shown to increase speech intelligibility in noise for cochlear implant (CI) listeners. It has been suggested that fundamental frequency (F0) provides this advantage. This study aimed to investigate the contribution of F0.MethodsSeven cochlear implant users having residual hearing at 125, 250, and 500 Hz contralateral to the implant were recruited. Speech intelligibility in noise was measured using an adaptive procedure for three listening conditions: (1) CI alone, (2) CI plus filtered acoustic information contralaterally, and (3) CI plus acoustic F0 contralaterally. In condition 2, the sentence material was low-passed at 500 Hz and presented via an insert earphone into the contralateral ear. In condition 3, F0 was extracted using Praat and presented as a sine wave with the same F0 variation over time as the original sentence.ResultsAlthough benefit was observed when low-frequency information was added for the majority of participants, on average no statistically significant difference was found for the three listening conditions.DiscussionThese results are not consistent with current literature. It is proposed that glimpsing cues may be responsible for the advantage observed in previous studies; in this study, both target and masker were presented in the acoustic condition and this may explain the discrepancy.ConclusionThe benefit of additional acoustic information may be highly variable and individual to participants, but on average no statistically significant difference was seen.