Monaural and binaural auditory frequency resolution measured using bandlimited noise and notched-noise masking

Several studies using bandlimited masking noise have indicated that NOSO frequency resolution is better than that for NOS pi. The present study examined NOSO and NOS pi frequency resolution with two different masking methods: bandlimited noise and notched noise. Noise spectrum levels of 10, 30, and 50 dB/Hz were used. Thresholds were determined for a 500-Hz signal, using a three-alternative forced-choice adaptive procedure, as a function of masker bandwidth and notchwidth. For NOSO presentation, 3-dB down points were comparable for the notched-noise and bandlimiting methods. For NOS pi presentation, 3-dB down points were generally greater for the bandlimiting method than the notched noise method. Furthermore, for NOS pi presentation, the 3-dB down estimate increased as noise level increased for the bandlimiting method, but stayed constant for the notched-noise method. It is suggested that the two masking methods measured different aspects of binaural processing.     

Auditory brain stem responses from human adults and infants: restriction of frequency contribution by notched-noise masking

The frequency contribution to the click-evoked ABR wave V was examined in adults and 3-month-old infants through the use of notch-filtered broadband noise. Notch center frequencies were set at 1.0, 4.0, and 8.0 kHz. Responses were obtained at 20, 40, and 60 dBnHL during the simultaneous presentation of each notched-noise masker as well as in an unmasked condition. The ABR wave V was analyzed for absolute latency and amplitude, as well as latency and amplitude changes resulting from the introduction of masking. Analyses showed wave V latency and amplitude values to be similar for adults and infants within the 1.0-kHz notch. Differences between adult and infant groups were observed as the notch was shifted to the high frequencies. Further, latency and amplitude shifts resulting from the introduction of masking noise produced differential effects on infant responses when compared to adults.     

Effects of signal delay on auditory filter shapes derived from psychophysical tuning curves and notched-noise data obtained in simultaneous masking

Psychophysical tuning curves (PTCs) measured in simultaneous masking usually sharpen as a short duration signal is moved from the onset to the temporal center of a longer duration masker. Filter shapes derived from notched-noise maskers have not consistently shown this effect. One possible explanation for this difference is that the signal level is fixed in the PTC paradigm, whereas the masker level is usually fixed in the notched-noise paradigm. In the present study, the signal level was fixed at 10 dB SL in both paradigms. The signal was 20 ms in duration, and presented at the onset or temporal center of the 400-ms masker. The masker was a pure tone presented in quiet (PTC) or in the presence of a pure-tone "restrictor" intended to limit off-frequency listening (PTCr), or it was a noise with a spectral notch placed symmetrically or asymmetrically about the 2-kHz signal frequency. Filter shapes were derived from the PTC, PTCr, and notched-noise data using the roex (p, w, t) model. The effects of signal delay and masking paradigm on filter bandwidth were analyzed with a two-factor repeated-measures ANOVA. There was a significant effect of signal delay (the filters sharpened with time) and masking paradigm (the filters derived from the notched-noise data were significantly wider than those derived from either of the PTC measurements, which did not differ from one another). Although the interaction between delay and paradigm was not significant, the filter derived from the notched-noise data sharpened more with time than did the other filters, and thus the bandwidth of the filters from the three paradigms were more similar at the longer delay than at the shorter delay. It is likely that the tuning-curve and notched-noise paradigms measure the same underlying filtering, but that various other factors contribute differentially to the derived filter shapes.     

Auditory filter shape derived from binaural masking experiments

The shape of the auditory filter was calculated from binaural masking experiments. Two different types of maskers were used in the study, a masker that was interaurally in phase at all frequencies (No), and a masker with an interaural phase difference of 0 below 500 Hz and of pi above 500 Hz. The test-signal frequency varied between 200 and 800 Hz, and the test signal was presented either monaurally (Sm) or binaurally in antiphase (S pi). By comparing the masked thresholds from the two experimental conditions, the following conclusion can be drawn: The threshold of the test signal is only affected by the masker phase within a narrow frequency range around the test frequency. Thus, for test-signal frequencies well above or below 500 Hz, no influence of the phase transition on the BMLD is observed, and normal masked thresholds for No and N pi maskers are obtained. For test frequencies around 500 Hz, the step in interaural phase difference leads to a decrease in the interaural correlation of the masker within the critical band around the test-signal frequency. This results in strong threshold changes for both monaural and binaural signals. A calculation of the auditory filter shape from the masked threshold values was performed under the assumption that the masked threshold is only dependent on the interaural cross correlation of the masker within the filter band. Using the formula of the EC theory for the relation between masker correlation and BMLD, the experimental data are well described by a trapezoidal filter with an equivalent rectangular bandwidth of 80 to 84 Hz.     

Auditory filter nonlinearity in mild/moderate hearing impairment

Sensorineural hearing loss has frequently been shown to result in a loss of frequency selectivity. Less is known about its effects on the level dependence of selectivity that is so prominent a feature of normal hearing. The aim of the present study is to characterize such changes in nonlinearity as manifested in the auditory filter shapes of listeners with mild/moderate hearing impairment. Notched-noise masked thresholds at 2 kHz were measured over a range of stimulus levels in hearing-impaired listeners with losses of 20-50 dB. Growth-of-masking functions for different notch widths are more parallel for hearing-impaired than for normal-hearing listeners, indicating a more linear filter. Level-dependent filter shapes estimated from the data show relatively little change in shape across level. The loss of nonlinearity is also evident in the input/output functions derived from the fitted filter shapes. Reductions in nonlinearity are clearly evident even in a listener with only 20-dB hearing loss.     

Additivity of simultaneous masking

Simultaneous masking functions (signal level at threshold versus masker level) were obtained for equally intense maskers presented individually and in pairs. The signal was a 2.0-kHz sinusoid. The pairs of maskers were (1) two sinusoids with frequencies 1.9 and 2.1 kHz, (2) two narrow bands of noise (50 Hz wide) centered at 1.9 and 2.1 kHz, (3) two narrow bands of noise (50 Hz wide) centered at 1.8 and 1.9 kHz, and (4) the 1.9-kHz sinusoid combined with the narrow band of noise centered at 2.1 kHz. The pairs of maskers produced anywhere from 10 to 17 dB of masking beyond that predicted from the simple sum of the masking produced by the individual maskers. The amount of this "additional masking" was independent of masker level. Adding a continuous low level background noise reduced the amount of additional masking only slightly (approximately 5 dB). The data are consistent with a model in which the effects of the maskers are summed after undergoing independent compressive transformations.     

Contributions of suppression and excitation to simultaneous masking: effects of signal frequency and masker-signal frequency relation

This study investigated the contributions of suppression and excitation to simultaneous masking for a range of masker frequencies both below and above three different signal frequencies (750, 2000, and 4850 Hz). A two-stage experiment was employed. In stage I, the level of each off-frequency simultaneous masker necessary to mask a signal at 10 or 30 dB sensation level was determined. In stage II, three different forward-masking conditions were tested: (1) an on-frequency condition, in which the signals in stage I were used to mask probes of the same frequency; (2) an off-frequency condition, in which the off-frequency maskers (at the levels determined in stage I) were used to mask the probes; and (3) a combined condition, in which the on- and off-frequency maskers were combined to mask the probes. If the off-frequency maskers simultaneously masked the signal via spread of excitation in stage I, then the off-frequency and combined maskers should produce considerable forward masking in stage II. If, on the other hand, they masked via suppression, they should produce little or no forward masking. The contribution of suppression was found to increase with increasing signal frequency; it was absent at 750 Hz, but dominant at 4850 Hz. These results have implications for excitation pattern analyses and are consistent with stronger nonlinear processing at high rather than at low frequencies.     

Transient masking and the temporal course of simultaneous tone-on-tone masking

Previous studies have shown that threshold for a signal in tone-on-tone simultaneous masking is sometimes lower when the masker is continuous than when it is gated. Threshold may also decline as signal onset is delayed relative to the onset of a longer duration masker, though it may increase again near masker offset. In the present study, the level of a 1250-Hz sinusoidal masker was found which would just mask a 20-ms, 1000-Hz sinusoid presented at 10-dB sensation level (SL). Masker duration was 20 or 400 ms; in the latter case, the signal was presented in one of three temporal positions within the masker. The level of the 1250-Hz masker necessary to mask the signal was reduced, sometimes by as much as 20-25 dB, by a 20-ms, 500-Hz sinusoid (transient masker) presented at the times when the signal might occur, but at a level 30 dB below that at which it would mask the 10-dB SL signal. This suggests that, in the earlier studies, at least some of the elevation in threshold in the presence of a short-duration masker or at the beginning (or end) of a longer duration masker may have been due to the transient responses to the masker affecting detection of the signal, but not necessarily masking the signal in terms of excitation in the signal "channel."     

Comparisons of frequency selectivity in simultaneous and forward masking for subjects with unilateral cochlear impairments

Two experiments are described in which frequency selectivity was estimated, in simultaneous and forward masking, for each ear of subjects with moderate (25-60 dB HL) unilateral cochlear hearing losses. In both experiments, the signal level was fixed for a given ear and type of masking (simultaneous or forward), and the masker level was varied to determine threshold, using an adaptive, two-alternative forced-choice procedure. In experiment I, the masker was a noise with a spectral notch centered at the signal frequency (either 1.0 or 1.5 kHz); threshold was determined as a function of notch width. Signal levels were chosen so that the noise level required at threshold for a notch width of zero was similar for the normal and impaired ear of each subject in both simultaneous and forward masking. The function relating threshold to notch width had a steeper slope for the normal ear than for the impaired ear of each subject. For the normal ears, these functions were steeper in forward masking than in simultaneous masking. This difference was interpreted as resulting from suppression. For the impaired ears, significant differences in the same direction were observed for three of the five subjects, but the differences were smaller. In experiment II, psychophysical tuning curves (PTCs) were determined in the presence of a fixed notched noise centered at the signal frequency (1.0 kHz). For the normal ears, the PTCs were sharper in forward masking than in simultaneous masking. For the impaired ears, the PTCs were similar in simultaneous and forward masking, but those in forward masking tended to be sharper at masker frequencies far removed from the signal frequency. Overall, the results suggest that suppression is reduced, but not completely absent in cases of moderate cochlear hearing loss.     

Additivity of simultaneous masking, revisited

Lutfi [J. Acoust. Soc. Am. 73, 262-267 (1983)] compared simultaneous masking functions (signal threshold versus masker level) for individual sinusoidal and narrow-band noise maskers, and for those maskers presented in pairs. Lutfi found that the pairs of maskers produced 10-17 dB "excess" masking over that predicted from the linear sum of their individual masking and explained the results in terms of a model in which the effects of the maskers are summed after undergoing independent compressive transformations. This paper describes experiments similar to those of Lutfi, and presents evidence suggesting that Lutfi's results may have been influenced by two factors: (1) combination-product detection, and (2) the use of different detection cues for single maskers and for pairs of maskers. Experiment I showed that when the stimulus conditions were chosen so as to minimize the likelihood of combination-product detection, "excess" masking was only 3-5 dB. Experiment II supported the idea that for a single narrow-band noise masker, subjects make use of the relatively slow envelope fluctuations to enhance performance. When two independent narrow-band noise maskers are added, the effectiveness of this cue is reduced, and between 3 and 9 dB of "excess" masking occurs. When the two noises are derived from the same source, and have correlated envelope fluctuations, no "excess" masking occurs. The results indicate that Lufti's compressive-nonlinearity model clearly fails in some situations.     

Auditory masking: need for improved conceptual structure

Even in simpler times, some people (e.g., Tanner) found it useful to ask "What is masking?" Independent of the extent to which this question was adequately answered even in those times, it is clear that the current expanded interest in central auditory processing has raised this question anew. In this note, comments are made about masking-related issues that illustrate the kinds of questions that need to be considered in attempting to develop a conceptual structure that can be effectively used to define, classify, study, and model auditory masking.     

联合深度编解码网络和时频掩蔽估计的单通道语音增强

提出了一种联合深度编解码神经网络和时频掩蔽估计的语音增强方法。该方法利用深度编解码网络估计时频掩蔽表示,并联合带噪语音的幅度谱学习带噪语音与纯净语音幅度谱之间的非线性映射关系。深度编解码网络采用卷积-反卷积网络结构。在编码端,利用卷积网络的局部感知特性,对带噪语音的时频域结构特征进行建模,提取语音特征,同时抑制背景噪声。在解码端,利用编码端提取到的语音特征逐层恢复局部细节信息并重构语音信号。同时,在编解码端对应层之间引入跳跃连接,以减少由于池化和全连接操作导致的低层细节信息丢失的问题。在TIMIT语音库和不完全匹配噪声集下进行仿真实验,实验结果表明,该方法可以有效抑制噪声,且能较好地恢复出语音细节成分。     

Growth of simultaneous masking for fm < fs: effects of overall frequency and level.

Growth-of-masking (GOM) functions were obtained in three groups of normal-hearing subjects using a simultaneous-masking paradigm. In all cases, the signal frequency (fs) was higher than the masker frequency (fm), either by a certain ratio (1.44) or by a certain distance [3 equivalent rectangular bandwidths (ERBs)]. The purpose was to evaluate the effect of overall frequency on the slope of the steep portion of the GOM function, and to evaluate the change in slope that can occur at high levels. Signal frequency ranged from 400 to 5000 Hz, and masker level ranged from 40 to 95 dB SPL. On average, the slope of the steep portion of the GOM function was about 1.4 for signal frequencies from 400 to 750 Hz, and 2.0 for signal frequencies from 1944 to 5000 Hz. This is consistent with the possibility that the cochlea may behave more linearly at the apical (low-frequency) region than at the basal (high-frequency) region. In addition, for signal frequencies at and above 750 Hz, the slope of the masking function changed from a value much greater than 1.0 to a value of 1.0 at high levels. The change in slope was better correlated with signal sensation level (i.e., amount of masking) than with either signal or masker SPL: the lack of a change at the lower signal frequencies may reflect the smaller amounts of masking there. The change to a linear growth of masking may represent a change in the response to the signal from compressive to linear at high levels.     

The temporal course of simultaneous tone-on-tone masking

Threshold for a 20-ms, 1-kHz signal was measured as a function of its temporal position within a longer duration gated masker; masker frequencies were below, at, and above 1 kHz. For a masker frequency above the signal frequency, there is a sizable temporal effect: As the onset of the signal is delayed, threshold decreases rapidly but then increases again as the signal approaches masker offset. Similar results can be observed for a masker frequency below the signal frequency, but that temporal effect is due to the detection of the cubic difference tone. The implication of this frequency-dependent temporal effect for measuring psychophysical tuning curves is discussed.     

Auditory filter shapes in the chinchilla

Auditory filter shapes were determined for the chinchilla using the notched-noise technique [R. D. Patterson, J. Acoust. Soc. Am. 59, 640-654 (1976)]. Here, the derivative of the curve relating threshold to masker gap width outlines the shape of the auditory filter. Three chinchillas were trained, using positive reinforcement techniques, to provide forward masked thresholds at 1.0 and 10.0 kHz, at three masker spectrum levels. Unexpectedly, the threshold curves contained inflection points and regions of constant or nonmonotonic changes in threshold, so that the derived filters contained dips in their central passbands. Nonmonotonic variations in threshold may be discerned in human threshold versus notch width functions of previously published studies, suggesting that the two types of data are qualitatively similar. The filters computed from the chinchilla data widened with increasing masker level and were more broadly tuned than those obtained in humans. The physiological response to each frequency component of any stimulus is likely a combination of excitation and suppression. Hence, one cannot predict masked threshold from the acoustic spectra of the maskers used here since they differ from their internal representations. Thus the threshold versus notch width function probably reflects the operation of both an auditory filter and a nonlinearity.     

中国人群的同时掩蔽测量及其特性分析

基于心理声学听音实验,进行了中国人群的听觉同时掩蔽阈值测量,其中掩蔽声为第4~22临界带带宽粉红噪声,掩蔽声压级60 dB和80 dB,被掩蔽声为频率在临界带频率范围内及其附近的纯音信号。在此基础上,对同时掩蔽特性进行了分析和讨论,并与相关文献的结论进行了比较。结果显示掩蔽曲线随着掩蔽声压级增大非对称性增强,上升沿与下降沿斜率差异增大。不同临界带的掩蔽曲线的平均峰值呈现无规分布,并不如前人模型所描述,从低频到高频单调衰减。掩蔽增长率在被掩蔽声频率较高时,普遍有一定的下降趋势,与前人测量结果有所差异。     

中国人群的同时掩蔽测量及其特性分析

基于心理声学听音实验,进行了中国人群的听觉同时掩蔽阈值测量,其中掩蔽声为第4~22临界带带宽粉红噪声,掩蔽声压级60 dB和80 dB,被掩蔽声为频率在临界带频率范围内及其附近的纯音信号。在此基础上,对同时掩蔽特性进行了分析和讨论,并与相关文献的结论进行了比较。结果显示掩蔽曲线随着掩蔽声压级增大非对称性增强,上升沿与下降沿斜率差异增大。不同临界带的掩蔽曲线的平均峰值呈现无规分布,并不如前人模型所描述,从低频到高频单调衰减。掩蔽增长率在被掩蔽声频率较高时,普遍有一定的下降趋势,与前人测量结果有所差异。