Number Of Maskers Research Articles

Selection of test-setup parameters to target specific signal-to-noise regions in speech-on-speech intelligibility testing

Objective: To investigate the magnitude of the change in speech-reception threshold (SRT) provided by altering four different test-setup parameters. Furthermore, to determine whether these changes in SRT are of a sufficient magnitude, such that they can be used to design a test-setup in future experiments that target a predefined signal-to-noise ratio (SNR) region. This could be particularly important if the test contrast investigated is confounded with test SNR. Design: The investigated test-setup parameters were: Spatial separation between target (0°) and maskers (±15°, ±30°, ±45°, or ±75°), number of maskers (two, four or six), scoring method (scoring percent-correct words or sentences) and masker gender (same or opposite to target). Twenty SRTs were measured per test subject. Study sample: Twenty hearing-impaired test subjects participated over two visits. Results: Alteration of masker gender, spatial separation between target and masker (±15°, ±30°, ± 45°), and scoring method was shown to offer SRT changes of a sufficient magnitude. The different test setups resulted in average SRTs ranging from −4.0 to 3.3 dB. Conclusion: Deliberately selecting test setup parameters can change the overall difficulty of the test by up to 7.3 dB SRT. Thus, a future experiment can, to this extent, be designed to target a specific SNR region.

Predicting binaural speech intelligibility using the signal-to-noise ratio in the envelope power spectrum domain.

This study proposes a binaural extension to the multi-resolution speech-based envelope power spectrum model (mr-sEPSM) [Jørgensen, Ewert, and Dau (2013). J. Acoust. Soc. Am. 134, 436-446]. It consists of a combination of better-ear (BE) and binaural unmasking processes, implemented as two monaural realizations of the mr-sEPSM combined with a short-term equalization-cancellation process, and uses the signal-to-noise ratio in the envelope domain (SNRenv) as the decision metric. The model requires only two parameters to be fitted per speech material and does not require an explicit frequency weighting. The model was validated against three data sets from the literature, which covered the following effects: the number of maskers, the masker types [speech-shaped noise (SSN), speech-modulated SSN, babble, and reversed speech], the masker(s) azimuths, reverberation on the target and masker, and the interaural time difference of the target and masker. The Pearson correlation coefficient between the simulated speech reception thresholds and the data across all experiments was 0.91. A model version that considered only BE processing performed similarly (correlation coefficient of 0.86) to the complete model, suggesting that BE processing could be considered sufficient to predict intelligibility in most realistic conditions.

Variations in the slope of the psychometric functions for speech intelligibility: a systematic survey.

Although many studies have looked at the effects of different listening conditions on the intelligibility of speech, their analyses have often concentrated on changes to a single value on the psychometric function, namely, the threshold. Far less commonly has the slope of the psychometric function, that is, the rate at which intelligibility changes with level, been considered. The slope of the function is crucial because it is the slope, rather than the threshold, that determines the improvement in intelligibility caused by any given improvement in signal-to-noise ratio by, for instance, a hearing aid. The aim of the current study was to systematically survey and reanalyze the psychometric function data available in the literature in an attempt to quantify the range of slope changes across studies and to identify listening conditions that affect the slope of the psychometric function. The data for 885 individual psychometric functions, taken from 139 different studies, were fitted with a common logistic equation from which the slope was calculated. Large variations in slope across studies were found, with slope values ranging from as shallow as 1% per dB to as steep as 44% per dB (median = 6.6% per dB), suggesting that the perceptual benefit offered by an improvement in signal-to-noise ratio depends greatly on listening environment. The type and number of maskers used were found to be major factors on the value of the slope of the psychometric function while other minor effects of target predictability, target corpus, and target/masker similarity were also found.

Predicting the speech reception thresholds with physical metrics

Many measures [e.g., speech transmission index (STI) and speech intelligibility index (SII)] have been proposed to predict the speech intelligibility in noise. Nevertheless, most of these studies were performed under the conditions with a limited number of maskers. The present study further investigated how well the present speech intelligibility and quality metrics predicted the speech reception thresholds (SRTs) for sentences corrupted by stationary and fluctuating maskers. The SRT scores were collected from 30 normal-hearing (NH) and 15 hearing-impaired (HI) native-Cantonese listeners. Sentences were corrupted by nine types of maskers, including speech-shaped noise and eight real-life environmental noises (4- and 6-talker babbles, upper and lower deck in bus, cafe, Chinese restaurant, MTR carriage, and street). The resulting average SRT scores were subject to the correlation analysis with various metrics computed from the noise-masked sentences. Of all the objective metrics considered, the STI and CSII measures performed the best, and their high correlations (i.e., r = 0.91 to 0.96) were maintained in both NH and HI conditions. This suggests that some of the physical metrics that have been found previously to correlate highly with the intelligibility of sentences in noise may also be used to predict the SRTs affected by different maskers.

Additivity of auditory masking using Gaussian-shaped tones

Both temporal and spectral masking have been studied extensively in the literature. Mostly, they have been regarded as separate phenomena. Very little is known about the interaction between these two effects, i.e. masking in the time-frequency domain. Data on the time-frequency spread of masking evoked by a single Gaussian-shaped tone pulse are presented in an accompanying study at the same conference (Necciari et al.). The current study gathers data on the additivity of masking by up to four, approximately equally effective Gaussian maskers (ERB=600 Hz), separated either along the time or the frequency axis. For temporal separation, the amount of masking increases with the number of maskers, with excess masking (exceeding linear additivity) of up to 25 dB. For frequency separation (preliminary data) excess masking amounts up to 15 dB, and the higher-frequency masker (relative to the target) contributes more to the additivity than the lower-frequency maskers. Experiments with multiple maskers combining both temporal and frequency separation are underway. Combined with the single maker data, these data may serve as a basis for modeling time-frequency masking effects in complex signals. Work partly supported by OEAD (WTZ project AMADEUS) and ANR

Informational masking as a function of masker set size

Thresholds for the detection of a 1000-Hz tone added to informational maskers were measured using six-tone maskers. A large number of different maskers were generated by randomly choosing the frequencies of the six tones from a (log) uniform distribution ranging from 100 to 6000 Hz, exempting the 160-Hz frequency range surrounding the 1000-Hz signal. These maskers were then assigned to different masker sets, with the number of maskers in any one set being 3, 6, 12, or 24. Thresholds were measured separately for the different masker sets. Unlike the method used by Wright and Saberi [J. Acoust. Soc. Am. 105, 1765–1775 (1999)], on each interval the masker was drawn with replacement. For small set sizes, the trials could be segregated into those in which the maskers were different across intervals and those in which the maskers were the same across intervals. The resulting thresholds did not depend on the masker composition in the two intervals. As the set size increases, the observers ability to remember/identify the maskers is diminished but the average thresholds are not obviously altered by changes in set size beyond 3. [Work supported by NIH.]

The role of auditory filters in comodulation masking release (CMR)

Comodulation masking release (CMR) is the detection advantage conferred by coherence of amplitude modulation across masker frequency. This phenomenon has typically been described in terms of across-frequency comparisons or in terms of cueing, where analysis of the output of an auditory filter in the region of the signal is aided by the outputs of independent auditory filters. An alternative approach is to assume a broad initial predetection filter, one which encompasses frequencies generally thought to fall into disparate auditory filters. These two basic approaches are compared. Stimuli consisted of comodulated maskers, spaced in frequency at one of three fixed intervals, with fine structure that either produced strong envelope beats in the summed waveform or did not. The signal was a pure tone of random starting phase. For the smallest masker frequency spacing detection of a high-frequency signal seemed to be improved in the presence of envelope beats, while detection of a low-frequency signal seemed to be degraded by envelope beats. These results are discussed in terms of the number of maskers assumed to sum in an auditory filter and the relative availability of within- and between-channel cues. This explanation is consistent with an initial stage of auditory filtering and is fundamentally inconsistent with a broad initial filter. Results for larger masker frequency spacings showed this trend less reliably, a finding that was further explored via data from a modulation discrimination task.

Effects of masker fluctuations, number of maskers, and reverberation on binaural speech perception in normal and impaired hearing

Binaural speech-reception thresholds (SRT) in noise for both normal-hearing and hearing-impaired listeners were determined (1) using earphone presentation of stimuli simulating a free-field environment with between one and six noise sources, and (2) in a reverberant environment with the listener in either the direct or indirect field of the speech and/or noise source. The masking noise was either steady state, or modulated like running speech. It was found that normal-hearing listeners benefit considerably from both masker fluctuations and interaural level differences due to headshadow: Both cues caused an advantage of up to 7 dB in terms of SRT. Much smaller gains were attained by most hearing impaired listeners. The gains also diminished when reverberation was present, or when the number of maskers was increased. Results further showed that in almost all conditions where interaural time or phase differences were present, the release from masking experienced by both normal-hearing and hearing-impaired listeners was in the order of 2–3 dB.

Forward masking of clicks with diotic and dichotic maskers

Thresholds for diotic clicks in a forward masking paradigm with broadband noise maskers were obtained as a function of noise level, temporal parameters of the masker, number of maskers and interaural phase of the masker (0° to 180°). Comparison of MLD's at two different masking intervals with masker intensities equated for amount of diotic masking, showed larger MLD's when the time between masker and signal was short. When two nonoverlapping maskers were combined, additional masking above that predicted by an intensity sum was obtained. The amount of additional masking in the dichotic condition was approximately equal to that in a diotic condition when the amount of masking produced by a single masker was equated in the diotic and dichotic conditions. Differences from previous studies that show additivity for the dichotic condition [K. Berg and W. A. Yost, J. Acoust. Soc. Am. 60, 173–177 (1976)] will be discussed. Within the context of an intensity coding model [M. J. Penner and R. M. Shiffrin, J. Acoust. Soc. Am. 67, 617 (1980)], the additional masking can be explained by an early compressive intensity transformation. [Work supported by NSF.]