Spectral Masking Research Articles

Perception of concurrent vowels: The effect of varying their relative level

When identifying pairs of simultaneous steady-state vowels, listeners perform well even when the two vowels start and stop at the same time, are presented monaurally, have the same fundamental frequency ( f 0), and have approximately equal intensities. The sensation described by listeners is of one dominant vowel “coloured” by a second or non-dominant vowel. A small difference in f 0 improves performance and typically results in a sensation of two voice sources rather than of one voice coloured by another. An experiment is reported in which four listeners attempted to identify separately the dominant and the non-dominant vowel when the relative levels of the vowels were varied over a 28 dB range and the vowels had either the same f 0 or different f 0s. As in previous experiments, performance improved when there was an f 0 difference between the vowels, but this advantage was reduced though not abolished when the f 0 separation was one octave. This advantage of f 0 separation was seen almost entirely in increased identification of the non-dominant vowel and was apparent at all but the greatest relative level differences tested. Finally, there were clear differences in the pattern of vowel dominance across the different vowel combinations, with subjects showing broadly similar patterns. Dominance may reflect cognitive “decision” strategies in addition to spectral masking at the level of the peripheral auditory system.

Analysis of picosecond pulse shape synthesis by spectral masking in a grating pulse compressor

This paper analyzes the synthesis of arbitrarily shaped optical pulses by spectral filtering in a fiber-and-grating pulse compressor. Spectral filtering of phase and amplitude is achieved by masking the spatially dispersed frequency components within the compressor. We show that the spectral filter is the convolution of the mask with the beam's transverse intensity profile. We discuss the effect of diffraction from the features of the physical mask, and show how finite spatial resolution limits the range of attainable temporal profiles. The fundamental limitation on spectral resolution is derived. Spectra and temporal pulse shapes corresponding to a variety of physical masks are calculated and are found to be in excellent agreement with experiments done with compressed pulses from a mode-locked Nd: YAG laser.

Speech competition effects on synthetic stop-vowel perception by normal and hearing-impaired listeners.

A triadic comparisons task and an identification task were used to evaluate normally hearing listeners' and hearing-impaired listeners' perceptions of synthetic CV stimuli in the presence of competition. The competing signals included multitalker babble, continuous speech spectrum noise, a CV masker, and a brief noise masker shaped to resemble the onset spectrum of the CV masker. All signals and maskers were presented monotically. Interference by competition was assessed by comparing Multidimensional Scaling solutions derived from each masking condition to that derived from the baseline (quiet) condition. Analysis of the effects of continuous maskers revealed that multitalker babble and continuous noise caused the same amount of change in performance, as compared to the baseline condition, for all listeners. CV masking changed performance significantly more than did brief noise masking, and the hearing-impaired listeners experienced more degradation in performance than normals. Finally, the velar CV maskers (g epsilon and k epsilon) caused significantly greater masking effects than the bilabial CV maskers (b epsilon and p epsilon), and were most resistant to masking by other competing stimuli. The results suggest that speech intelligibility difficulties in the presence of competing segments of speech are primarily attributable to phonetic interference rather than to spectral masking. Individual differences in hearing-impaired listeners' performances are also discussed.