Effects of spectro-temporal modulation changes produced by multi-channel compression on intelligibility in a competing-speech task

These experiments are concerned with the intelligibility of target speech in the presence of a background talker. Using a noise vocoder, Stone and Moore [J. Acoust. Soc. Am. 114, 1023-1034 (2003)] showed that single-channel fast-acting compression degraded intelligibility, but slow compression did not. Stone and Moore [J. Acoust. Soc. Am. 116, 2311-2323 (2004)] showed that intelligibility was lower when fast single-channel compression was applied to the target and background after mixing rather than before, and suggested that this was partly due to compression after mixing introducing "comodulation" between the target and background talkers. Experiment 1 here showed a similar effect for multi-channel compression. In experiment 2, intelligibility was measured as a function of the speed of multi-channel compression applied after mixing. For both eight- and 12-channel vocoders with one compressor per channel, intelligibility decreased as compression speed increased. For the eight-channel vocoder, a compressor that only affected modulation depth for rates below 2 Hz still reduced intelligibility. Experiment 3 used 12- or 18-channel vocoders. There were between 1 and 12 compression channels, and four speeds of compression. Intelligibility decreased as the number and speed of compression channels increased. The results are interpreted using several measures of the effects of compression, especially "across-source modulation correlation."     

Channel selection in the modulation domain for improved speech intelligibility in noise

Background noise reduces the depth of the low-frequency envelope modulations known to be important for speech intelligibility. The relative strength of the target and masker envelope modulations can be quantified using a modulation signal-to-noise ratio, (S/N)(mod), measure. Such a measure can be used in noise-suppression algorithms to extract target-relevant modulations from the corrupted (target + masker) envelopes for potential improvement in speech intelligibility. In the present study, envelopes are decomposed in the modulation spectral domain into a number of channels spanning the range of 0-30 Hz. Target-dominant modulations are identified and retained in each channel based on the (S/N)(mod) selection criterion, while modulations which potentially interfere with perception of the target (i.e., those dominated by the masker) are discarded. The impact of modulation-selective processing on the speech-reception threshold for sentences in noise is assessed with normal-hearing listeners. Results indicate that the intelligibility of noise-masked speech can be improved by as much as 13 dB when preserving target-dominant modulations, present up to a modulation frequency of 18 Hz, while discarding masker-dominant modulations from the mixture envelopes.     

Predicting the intelligibility of vocoded and wideband Mandarin Chinese

Due to the limited number of cochlear implantees speaking Mandarin Chinese, it is extremely difficult to evaluate new speech coding algorithms designed for tonal languages. Access to an intelligibility index that could reliably predict the intelligibility of vocoded (and non-vocoded) Mandarin Chinese is a viable solution to address this challenge. The speech-transmission index (STI) and coherence-based intelligibility measures, among others, have been examined extensively for predicting the intelligibility of English speech but have not been evaluated for vocoded or wideband (non-vocoded) Mandarin speech despite the perceptual differences between the two languages. The results indicated that the coherence-based measures seem to be influenced by the characteristics of the spoken language. The highest correlation (r = 0.91-0.97) was obtained in Mandarin Chinese with a weighted coherence measure that included primarily information from high-intensity voiced segments (e.g., vowels) containing F0 information, known to be important for lexical tone recognition. In contrast, in English, highest correlation was obtained with a coherence measure that included information from weak consonants and vowel/consonant transitions. A band-importance function was proposed that captured information about the amplitude envelope contour. A higher modulation rate (100 Hz) was found necessary for the STI-based measures for maximum correlation (r = 0.94-0.96) with vocoded Mandarin and English recognition.     

Predicted effects of sensorineural hearing loss on across-fiber envelope coding in the auditory nerve

Cross-channel envelope correlations are hypothesized to influence speech intelligibility, particularly in adverse conditions. Acoustic analyses suggest speech envelope correlations differ for syllabic and phonemic ranges of modulation frequency. The influence of cochlear filtering was examined here by predicting cross-channel envelope correlations in different speech modulation ranges for normal and impaired auditory-nerve (AN) responses. Neural cross-correlation coefficients quantified across-fiber envelope coding in syllabic (0-5 Hz), phonemic (5-64 Hz), and periodicity (64-300 Hz) modulation ranges. Spike trains were generated from a physiologically based AN model. Correlations were also computed using the model with selective hair-cell damage. Neural predictions revealed that envelope cross-correlation decreased with increased characteristic-frequency separation for all modulation ranges (with greater syllabic-envelope correlation than phonemic or periodicity). Syllabic envelope was highly correlated across many spectral channels, whereas phonemic and periodicity envelopes were correlated mainly between adjacent channels. Outer-hair-cell impairment increased the degree of cross-channel correlation for phonemic and periodicity ranges for speech in quiet and in noise, thereby reducing the number of independent neural information channels for envelope coding. In contrast, outer-hair-cell impairment was predicted to decrease cross-channel correlation for syllabic envelopes in noise, which may partially account for the reduced ability of hearing-impaired listeners to segregate speech in complex backgrounds.     

A detailed study on the effects of noise on speech intelligibility

A wavelet representation of speech was used to display the instantaneous amplitude and phase within 14 octave frequency bands, representing the envelope and the carrier within each band. Adding stationary noise alters the wavelet pattern, which can be understood as a combination of three simultaneously occurring subeffects: two effects on the wavelet levels (one systematic and one stochastic) and one effect on the wavelet phases. Specific types of signal processing were applied to speech, which allowed each effect to be either included or excluded. The impact of each effect (and of combinations) on speech intelligibility was measured with CVC's. It appeared that the systematic level effect (i.e., the increase of each speech wavelet intensity with the mean noise intensity) has the most degrading effect on speech intelligibility, which is in accordance with measures such as the modulation transfer function and the speech transmission index. However, also the introduction of stochastic level fluctuations and disturbance of the carrier phase seriously contribute to reduced intelligibility in noise. It is argued that these stochastic effects are responsible for the limited success of spectral subtraction as a means to improve speech intelligibility. Results can provide clues for effective noise suppression with respect to intelligibility.     

Predicting speech intelligibility based on the signal-to-noise envelope power ratio after modulation-frequency selective processing

A model for predicting the intelligibility of processed noisy speech is proposed. The speech-based envelope power spectrum model has a similar structure as the model of Ewert and Dau [(2000). J. Acoust. Soc. Am. 108, 1181-1196], developed to account for modulation detection and masking data. The model estimates the speech-to-noise envelope power ratio, SNR(env), at the output of a modulation filterbank and relates this metric to speech intelligibility using the concept of an ideal observer. Predictions were compared to data on the intelligibility of speech presented in stationary speech-shaped noise. The model was further tested in conditions with noisy speech subjected to reverberation and spectral subtraction. Good agreement between predictions and data was found in all cases. For spectral subtraction, an analysis of the model's internal representation of the stimuli revealed that the predicted decrease of intelligibility was caused by the estimated noise envelope power exceeding that of the speech. The classical concept of the speech transmission index fails in this condition. The results strongly suggest that the signal-to-noise ratio at the output of a modulation frequency selective process provides a key measure of speech intelligibility.     

Effect of the speed of a single-channel dynamic range compressor on intelligibility in a competing speech task

Using a "noise-vocoder" cochlear implant simulator [Shannon et al., Science 270, 303-304 (1995)], the effect of the speed of dynamic range compression on speech intelligibility was assessed, using normal-hearing subjects. The target speech had a level 5 dB above that of the competing speech. Initially, baseline performance was measured with no compression active, using between 4 and 16 processing channels. Then, performance was measured using a fast-acting compressor and a slow-acting compressor, each operating prior to the vocoder simulation. The fast system produced significant gain variation over syllabic timescales. The slow system produced significant gain variation only over the timescale of sentences. With no compression active, about six channels were necessary to achieve 50% correct identification of words in sentences. Sixteen channels produced near-maximum performance. Slow-acting compression produced no significant degradation relative to the baseline. However, fast-acting compression consistently reduced performance relative to that for the baseline, over a wide range of performance levels. It is suggested that fast-acting compression degrades performance for two reasons: (1) because it introduces correlated fluctuations in amplitude in different frequency bands, which tends to produce perceptual fusion of the target and background sounds and (2) because it reduces amplitude modulation depth and intensity contrasts.     

Temporal envelope compensation for robust phoneme recognition using modulation spectrum

A robust feature extraction technique for phoneme recognition is proposed which is based on deriving modulation frequency components from the speech signal. The modulation frequency components are computed from syllable-length segments of sub-band temporal envelopes estimated using frequency domain linear prediction. Although the baseline features provide good performance in clean conditions, the performance degrades significantly in noisy conditions. In this paper, a technique for noise compensation is proposed where an estimate of the noise envelope is subtracted from the noisy speech envelope. The noise compensation technique suppresses the effect of additive noise in speech. The robustness of the proposed features is further enhanced by the gain normalization technique. The normalized temporal envelopes are compressed with static (logarithmic) and dynamic (adaptive loops) compression and are converted into modulation frequency features. These features are used in an automatic phoneme recognition task. Experiments are performed in mismatched train/test conditions where the test data are corrupted with various environmental distortions like telephone channel noise, additive noise, and room reverberation. Experiments are also performed on large amounts of real conversational telephone speech. In these experiments, the proposed features show substantial improvements in phoneme recognition rates compared to other speech analysis techniques. Furthermore, the contribution of various processing stages for robust speech signal representation is analyzed.     

Consonant identification under maskers with sinusoidal modulation: masking release or modulation interference?

The present study investigated the effect of envelope modulations in a background masker on consonant recognition by normal hearing listeners. It is well known that listeners understand speech better under a temporally modulated masker than under a steady masker at the same level, due to masking release. The possibility of an opposite phenomenon, modulation interference, whereby speech recognition could be degraded by a modulated masker due to interference with auditory processing of the speech envelope, was hypothesized and tested under various speech and masker conditions. It was of interest whether modulation interference for speech perception, if it were observed, could be predicted by modulation masking, as found in psychoacoustic studies using nonspeech stimuli. Results revealed that masking release measurably occurred under a variety of conditions, especially when the speech signal maintained a high degree of redundancy across several frequency bands. Modulation interference was also clearly observed under several circumstances when the speech signal did not contain a high redundancy. However, the effect of modulation interference did not follow the expected pattern from psychoacoustic modulation masking results. In conclusion, (1) both factors, modulation interference and masking release, should be accounted for whenever a background masker contains temporal fluctuations, and (2) caution needs to be taken when psychoacoustic theory on modulation masking is applied to speech recognition.     

Principal-component amplitude compression for the hearing impaired

Principal-component amplitude compression, a means for matching speech to the reduced dynamic range in sensorineural hearing impairments, is a multiband approach aimed at preserving details of spectral shape while reducing overall level variation. The effect of compression has been studied for the first and second principal components (PC1 an PC2) of the short-term speech spectrum, which are roughly representative of overall level and spectral tilt, respectively. Compression of PC1 roughly equalizes consonant and vowel levels while compression of PC2 provides time-varying high-frequency emphasis. The effect on speech intelligibility of sensorineural hearing-impaired listeners of two principal-component compression system implementations, compression of PC1 and compression of both PC1 and PC2, was compared to that of linear amplification (LA), independent compression of multiple bands (MBC), and wideband compression (WC). Results indicate that compression of overall level as provided by compression of PC1 and WC improved intelligibility relative to LA over a 10- to 15-dB range of input levels. While MBC was beneficial in some cases, it did not provide higher intelligibility than WC. Compression of PC2 did not benefit but rather degraded performance relative to LA. Error analyses and band-level measurements indicate that the highest intelligibility is obtained when audibility is improved and the relative spectral shapes of different speech sounds are preserved.     

Coherence depression in stochastic excitable systems with two-frequency forcing

We study the response of two generic neuron models, the leaky integrate-and-fire (LIF) model and the leaky integrate-and-fire model with dynamic threshold (LIFDT) (i.e., with memory) to a stimulus consisting of two sinusoidal drives with incommensurate frequency, an amplitude modulation ("envelope") noise and a relatively weak additive noise. Spectral and coherence analysis of responses to such naturalistic stimuli reveals how the LIFDT model exhibits better correlation between modulation and spike train even in the presence of both noises. However, a resonance-induced synchrony, occurring when the beat frequency between the sinusoids is close to the intrinsic neuronal firing rate, decreases the coherence in the dynamic threshold case. Under suprathreshold conditions, the modulation noise simultaneously decreases the linear spectral coherence between the spikes and the whole stimulus, as well as between spikes and the stimulus envelope. Our study shows that the coefficient of variation of the envelope fluctuations is positively correlated with the degree of coherence depression. As the coherence function quantifies the linear information transmission, our findings indicate that under certain conditions, a transmission loss results when an excitable system with adaptive properties encodes a beat with frequency in the vicinity of its mean firing rate.     

Evaluation of companding-based spectral enhancement using simulated cochlear-implant processing

This study tested a time-domain spectral enhancement algorithm that was recently proposed by Turicchia and Sarpeshkar [IEEE Trans. Speech Audio Proc. 13, 243-253 (2005)]. The algorithm uses a filter bank, with each filter channel comprising broadly tuned amplitude compression, followed by more narrowly tuned expansion (companding). Normal-hearing listeners were tested in their ability to recognize sentences processed through a noise-excited envelope vocoder that simulates aspects of cochlear-implant processing. The sentences were presented in a steady background noise at signal-to-noise ratios of 0, 3, and 6 dB and were either passed directly through an envelope vocoder, or were first processed by the companding algorithm. Using an eight-channel envelope vocoder, companding produced small but significant improvements in speech reception. Parametric variations of the companding algorithm showed that the improvement in intelligibility was robust to changes in filter tuning, whereas decreases in the time constants resulted in a decrease in intelligibility. Companding continued to provide a benefit when the number of vocoder frequency channels was increased to sixteen. When integrated within a sixteen-channel cochlear-implant simulator, companding also led to significant improvements in sentence recognition. Thus, companding may represent a readily implementable way to provide some speech recognition benefits to current cochlear-implant users.     

Effects of periodic interruptions on the intelligibility of speech based on temporal fine-structure or envelope cues

The intelligibility of speech signals processed to retain either temporal envelope (E) or fine structure (TFS) cues within 16 0.4-oct-wide frequency bands was evaluated when processed stimuli were periodically interrupted at different rates. The interrupted E- and TFS-coded stimuli were highly intelligible in all conditions. However, the different patterns of results obtained for E- and TFS-coded speech suggest that the two types of stimuli do not convey identical speech cues. When an effect of interruption rate was observed, the effect occurred at low interruption rates (<8 Hz) and was stronger for E- than TFS-coded speech, suggesting larger involvement of modulation masking with E-coded speech.     

Speech segregation in rooms: monaural, binaural, and interacting effects of reverberation on target and interferer

Speech reception thresholds were measured in virtual rooms to investigate the influence of reverberation on speech intelligibility for spatially separated targets and interferers. The measurements were realized under headphones, using target sentences and noise or two-voice interferers. The room simulation allowed variation of the absorption coefficient of the room surfaces independently for target and interferer. The direct-to-reverberant ratio and interaural coherence of sources were also varied independently by considering binaural and diotic listening. The main effect of reverberation on the interferer was binaural and mediated by the coherence, in agreement with binaural unmasking theories. It appeared at lower reverberation levels than the effect of reverberation on the target, which was mainly monaural and associated with the direct-to-reverberant ratio, and could be explained by the loss of amplitude modulation in the reverberant speech signals. This effect was slightly smaller when listening binaurally. Reverberation might also be responsible for a disruption of the mechanism by which the auditory system exploits fundamental frequency differences to segregate competing voices, and a disruption of the "listening in the gaps" associated with speech interferers. These disruptions may explain an interaction observed between the effects of reverberation on the targets and two-voice interferers.     

汉语音段反转言语的可懂度研究

实验研究了帧长对汉语音段反转言语可懂度的影响。实验结果表明,帧长在64 ms以下,汉语音段反转言语具有较高的可懂度;帧长在64~203 ms之间,可懂度随帧长的增加逐渐降低;帧长在203 ms以上,可懂度为0。在帧长8 ms时,汉语的声调失真导致可懂度下降。原始语音信号和音段反转言语的调制谱的分析表明,调制谱失真大小和可懂度密切相关。因此,用原始语音信号和音段反转言语的窄带包络间的归一化相关值可以衡量调制谱失真大小,基于语音的语言传输指数法计算的客观值和实验结果显著相关(r=0.876,p<0.01)。研究表明,语言可懂度与窄带包络有关,音段反转言语的可懂度和保留原始语音信号的窄带包络密切相关。      

The effects of the addition of low-level, low-noise noise on the intelligibility of sentences processed to remove temporal envelope information

The intelligibility of sentences processed to remove temporal envelope information, as far as possible, was assessed. Sentences were filtered into N analysis channels, and each channel signal was divided by its Hilbert envelope to remove envelope information but leave temporal fine structure (TFS) intact. Channel signals were combined to give TFS speech. The effect of adding low-level low-noise noise (LNN) to each channel signal before processing was assessed. The addition of LNN reduced the amplification of low-level signal portions that contained large excursions in instantaneous frequency, and improved the intelligibility of simple TFS speech sentences, but not more complex sentences. It also reduced the time needed to reach a stable level of performance. The recovery of envelope cues by peripheral auditory filtering was investigated by measuring the intelligibility of 'recovered-envelope speech', formed by filtering TFS speech with an array of simulated auditory filters, and using the envelopes at the output of these filters to modulate sinusoids with frequencies equal to the filter center frequencies (i.e., tone vocoding). The intelligibility of TFS speech and recovered-envelope speech fell as N increased, although TFS speech was still highly intelligible for values of N for which the intelligibility of recovered-envelope speech was low.     

Quantifying the intelligibility of speech in noise for non-native talkers

The intelligibility of speech pronounced by non-native talkers is generally lower than speech pronounced by native talkers, especially under adverse conditions, such as high levels of background noise. The effect of foreign accent on speech intelligibility was investigated quantitatively through a series of experiments involving voices of 15 talkers, differing in language background, age of second-language (L2) acquisition and experience with the target language (Dutch). Overall speech intelligibility of L2 talkers in noise is predicted with a reasonable accuracy from accent ratings by native listeners, as well as from the self-ratings for proficiency of L2 talkers. For non-native speech, unlike native speech, the intelligibility of short messages (sentences) cannot be fully predicted by phoneme-based intelligibility tests. Although incorrect recognition of specific phonemes certainly occurs as a result of foreign accent, the effect of reduced phoneme recognition on the intelligibility of sentences may range from severe to virtually absent, depending on (for instance) the speech-to-noise ratio. Objective acoustic-phonetic analyses of accented speech were also carried out, but satisfactory overall predictions of speech intelligibility could not be obtained with relatively simple acoustic-phonetic measures.     

Influences of auditory object formation on phonemic restoration

In phonemic restoration, intelligibility of interrupted speech is enhanced when noise fills the speech gaps. When the broadband envelope of missing speech amplitude modulates the intervening noise, intelligibility is even better. However, this phenomenon represents a perceptual failure: The amplitude modulation, a noise feature, is misattributed to the speech. Experiments explored whether object formation influences how information in the speech gaps is perceptually allocated. Experiment 1 replicates the finding that intelligibility is enhanced when speech-modulated noise rather than unmodulated noise is presented in the gaps. In Experiment 2, interrupted speech was presented diotically, but intervening noises were presented either diotically or with an interaural time difference leading in the right ear, causing the noises to be perceived to the side of the listener. When speech-modulated noise and speech are perceived from different directions, intelligibility is no longer enhanced by the modulation. However, perceived location has no effect for unmodulated noise, which contains no speech-derived information. Results suggest that enhancing object formation reduces misallocation of acoustic features across objects, and demonstrate that our ability to understand noisy speech depends on a cascade of interacting processes, including glimpsing sensory inputs, grouping sensory inputs into objects, and resolving ambiguity through top-down knowledge.     

Improvement of intelligibility of ideal binary-masked noisy speech by adding background noise

When a target-speech/masker mixture is processed with the signal-separation technique, ideal binary mask (IBM), intelligibility of target speech is remarkably improved in both normal-hearing listeners and hearing-impaired listeners. Intelligibility of speech can also be improved by filling in speech gaps with un-modulated broadband noise. This study investigated whether intelligibility of target speech in the IBM-treated target-speech/masker mixture can be further improved by adding a broadband-noise background. The results of this study show that following the IBM manipulation, which remarkably released target speech from speech-spectrum noise, foreign-speech, or native-speech masking (experiment 1), adding a broadband-noise background with the signal-to-noise ratio no less than 4 dB significantly improved intelligibility of target speech when the masker was either noise (experiment 2) or speech (experiment 3). The results suggest that since adding the noise background shallows the areas of silence in the time-frequency domain of the IBM-treated target-speech/masker mixture, the abruption of transient changes in the mixture is smoothed and the perceived continuity of target-speech components becomes enhanced, leading to improved target-speech intelligibility. The findings are useful for advancing computational auditory scene analysis, hearing-aid/cochlear-implant designs, and understanding of speech perception under "cocktail-party" conditions.