Spectral Smearing Research Articles

Bat jitter discrimination, adaptive channel equalization, and the critical interval

When simulated echoes have a bandwidth between 20 and 80 kHz, bats (Eptesicus fuscus) can discriminate between pulse-to-pulse delay jitter and constant echo delay with a sensitivity of 10 ns [Simmons et al., J. Comp. Physiol. A 167, 589–616 (1990)]. To investigate this amazing acuity, high-pass and low-pass filters were inserted into the channel between the simulated target and the bat. The effect of such filters on jitter acuity was measured. The data are explained by a receiver model that attempts to equalize the channel filter. The equalizer uses a smeared version of the actual channel transfer function, where spectral smearing induces relatively slow cutoff. Spectral smearing occurs in spectrogram (short-time Fourier transform) analysis. Smearing varies inversely with the impulse response duration of the analysis filters. This impulse response duration corresponds to the critical interval, which is frequency dependent for proportional bandwidth filtering. Bat jitter discrimination data imply channel equalization with a smeared version of the actual channel transfer function. This smearing is commensurate with critical intervals that are obtained from other experiments [Simmons et al., J. Acoust. Soc. Am. 86, 1318–1332 (1989)]. Bats apparently use time-frequency echo representations to perform environmentally adaptive filtering.

Speech recognition with altered spectral distribution of envelope cues

Recognition of consonants, vowels, and sentences was measured in conditions of reduced and altered spectral information. Speech materials were passed through four bandpass filters (analysis bands). The output of each band was half-wave rectified, and low-pass filtered at 160 Hz to extract the temporal envelope. The envelope from each band modulated a bandlimited noise (carrier bands). Experiment I explored the importance of the cutoff frequencies of the bands, which proved not to be a critical parameter for speech recognition as long as the analysis and carrier bands were the same. Experiment II warped the spectral distribution of envelope cues by mismatching the analysis and carrier bands in frequency extent, resulting in a dramatic decrease in performance. Experiment III simulated the spectral smearing caused by broad auditory filters or electrode interaction in a cochlear implant by decreasing the slopes on the noise carrier bands. A decrease in performance was observed only for the most broadly overlapping carrier bands, indicating that speech pattern recognition is robust to spectral smearing. Overall, these results show that, for four bands, the most critical parameter is the frequency alignment of the analysis bands and carrier bands. [Supported by NIDCD.]

Effects of spectral smearing on phoneme and word recognition.

The principal goal was to measure the effects, on speech perception, of loss of spectral detail in the acoustic signal. Spectral smearing was produced by multiplying the speech waveform by low-pass filtered noise. Performance was measured in normal adults as the percentage of phonemes correctly repeated in lists of monosyllabic words. A smearing bandwidth of 250 Hz (i.e., each tonal component of the instantaneous spectrum replaced by a 250-Hz band of noise) had a small but significant effect on phoneme recognition. A smearing bandwidth of 8000 Hz was required to reduce phoneme recognition to a value that was indistinguishable from that produced by complete smearing. Vowels were somewhat more susceptible to the effects of spectral smearing than were consonants, and initial consonants were more susceptible than were final consonants. In an analysis of errors, place of consonant articulation was more susceptible than either manner of articulation or voicing. These findings are attributable to differences in the relative importance of spectral and temporal cues. Word recognition was more susceptible to the effects of spectral smearing than was phoneme recognition, but this finding was predictable on the basis of the known nonlinear relationship between the two measures. In a second experiment, smearing bandwidths of 707 and 2000 Hz increased phoneme recognition threshold by 12.9 and 16.4 dB, respectively, compared to that found without smearing. (Phoneme recognition threshold is defined, here, as the signal-to-noise ratio at which phoneme recognition is 50% of the value obtained in quiet.) The data are consistent with the hypothesis that reduced spectral resolution affects phoneme recognition to the extent that it reduces access to the formant patterns in the spectral envelope.

Simulation of sensorineural hearing loss: Use of linear spectral smearing to reduce frequency resolution

Phoneme and word recognition, was measured in seven hearing adults under various conditions of spectral smearing—produced by multiplying the speech waveform by low‐pass‐filtered noise. Phoneme recognition fell from 97% for no smearing to 13% for complete smearing (±10 kHz). Smearing by ±600 Hz reduced group mean phoneme recognition to 50%—a score typical of hearing‐impaired subjects with hearing losses around 80 dB, listening to amplified speech. In a second experiment, noise was mixed with the speech before spectral smearing was introduced. Noise susceptibility was measured as the S/N ratio required for a phoneme recognition score that was 50% of the score in quiet. Smearing at ±1 kHz increased noise susceptibility by 20 dB, an amount similar to that found in hearing‐impaired subjects with losses in excess of 90 dB. As with hearing‐impaired subjects, spectral smearing diminished the perception of consonant place more than the perception of consonant manner and voicing. Unlike with hearing‐impaired subjects, however, smearing affected vowel perception more than consonant perception. This last finding may be attributed to the use smearing over a fixed bandwidth—producing a greater percentage smearing in the lower frequencies. [Work supported by NIDCD Grant No. 10078.]

Spectral enhancement to compensate for reduced frequency selectivity

A new method of spectral enhancement has been developed to compensate for reduced frequency selectivity in people with sensorineural hearing impairment. The goal is to modify a sound such that it evokes as nearly as possible the same excitation patterns in an ear with reduced frequency selectivity as would be evoked by the unmodified sound in a normal ear. Processing is performed in the spectral domain, using an overlap-add procedure. Enhanced spectra are calculated using a nonnegative-least-squares algorithm for matching excitation patterns. When compensating for a moderate degree of auditory-fiber broadening, this procedure dramatically increases spectral contrast while preserving relatively normal sound quality. To examine the effects of enhancement on the intelligibility of speech in speech-shaped noise (at −3 dB SNR), a pilot experiment was performed with normal-hearing subjects, using a spectral smearing method [T. Baer and B. C. J. Moore, J. Acoust. Soc. Am. 94, 1229–1241 (1993)] to simulate some aspects of reduced frequency selectivity. Results show that the enhancement compensates partially for the effects of spectral smearing and should thus improve the intelligibilty of speech in noise for the hearing-impaired. Experiments are presently underway to test the effects of enhancement on speech perception with hearing-impaired subjects. [Work supported by the MRC.]

Effects of spectral smearing on the intelligibility of sentences in the presence of interfering speech

In a previous study [T. Baer and B. C. J. Moore, J. Acoust. Soc. Am. 94, 1229–1241 (1993)], a spectral smearing technique was used to simulate some of the effects of impaired frequency selectivity so as to assess its influence on speech intelligibility. Results showed that spectral smearing to simulate broadening of the auditory filters by a factor of 3 or 6 had little effect on the intelligibility of speech in quiet but had a large effect on the intelligibility of speech in noise. The present study examines the effect of spectral smearing on the intelligibility of speech in the presence of a single interfering talker. The results were generally consistent with those of the previous study, suggesting that impaired frequency selectivity contributes significantly to the problems experienced by people with cochlear hearing loss when they listen to speech in the presence of interfering sounds.

Effects of spectral smearing on the intelligibility of sentences in noise

Cochlear hearing loss is generally associated with a variety of deficits in the ability to analyze sounds, including reduced frequency selectivity. The complex interrelationships between the different deficits have made it difficult to assess the contribution of reduced frequency selectivity to the difficulties experienced by hearing-impaired people in understanding speech. To address this question, the effects of reduced frequency selectivity on the place representation of speech stimuli in the auditory system were simulated by ‘‘smearing’’ the spectra of the stimuli, using the overlap–add method. The smearing was designed to evoke excitation patterns in a normal ear that would resemble those evoked in an impaired ear using unsmeared stimuli. Several different types of smearing were used, simulating specific degrees of broadening and asymmetry of the auditory filter. Sentences in quiet and in speech-shaped noise were smeared and presented to normally hearing listeners in intelligibility tests. The results showed that the intelligibility of speech in quiet was hardly affected by spectral smearing, even for smearing that simulated auditory filters six times broader than normal. However, the intelligibility of speech in noise was adversely affected by the smearing, especially for large degrees of smearing and at a low speech-to-noise ratio (−3 dB). Simulation of asymmetrical broadening of the lower side of the auditory filter had a greater effect than simulation of asymmetrical broadening of the upper side, suggesting that upward spread of masking may be particularly important.

The influence of frequency resolution on the detection of spectral contrast by hearing‐impaired listeners

Abnormal frequency resolution associated with sensorineural hearing impairment produces a smearing of spectral detail in the internal representation of complex acoustic stimuli. As a result, listeners with hearing loss may have difficulty locating spectral peaks that provide important cues to speech understanding. This study examined the relationship between frequency separation of peaks in a complex sound and the degree of spectral contrast preserved in normal and impaired auditory systems. Five hearing‐impaired subjects (HI) and five normal‐hearing subjects (NH) discriminated a flat‐spectrum bandpass stimulus from a stimulus containing a sinusoidal ripple across its frequency range. The peak‐to‐valley amplitude (in dB) necessary for detection of the ripple was measured for ripple frequencies ranging from 1 to 9 cycles/oct. Auditory filter characteristics in notched‐noise were measured at 1 and 3 kHz in order to assess the relationship between frequency resolution and the ability to detect spectral contrast in complex spectra. There were clear differences between groups in both auditory filter characteristics and spectral contrast detection. Auditory filters tended to be both broader and more asymmetric for HI listeners than for NH listeners. Mean ripple amplitudes at threshold were approximately 4 dB lower for the NH group than the HI group at all ripple frequencies. However, excitation patterns based on auditory filter characteristics and threshold stimuli specific to individual listeners exhibited levels of peak‐to‐valley contrast that were nearly identical across listeners. This suggests that increased ripple detection thresholds of the hearing‐impaired listeners were due to spectral smearing resulting from impaired frequency resolution. [Work supported by NIDCD Grant No. DC00626.]

The internal signal-to-noise ratio in hearing-impaired listeners

The reduced frequency resolution often demonstrated by listeners with sensorineural hearing impairment might have at least two related effects on their perception of speech in background noise. The internal representation of the peaks and valleys in the speech spectrum become smeared, resulting in less precise frequency analysis than can be achieved by normal-hearing persons. In addition, when the speech is embedded in noise, poorer frequency resolution will, in theory, result in a greater-than-normal decrease in the signal-to-noise ratio at the outputs of the auditory filters. This poorer-than-normal internal S/N might account for the common observation that noise is more degrading to speech understanding by hearing-impaired listeners than by normal-hearing listeners. To assess the validity of this notion, both normal-hearing and hearing-impaired subjects participated in an identification experiment using highly simplified vowel-like harmonic complexes presented in quiet and in three levels of broadband noise. After equating individual performance in terms of percent correct identification in quiet, trading relationships were established between amount of spectral contrast in the stimuli (level differences between peaks and valleys in the spectra) and levels of the background noise. When the effect of spectral smearing introduced by reduced frequency analysis was compensated by increasing the input spectral contrast, the effects of additional background noise were similar for normal-hearing and hearing-impaired groups of subjects. Results from individual listeners will be discussed in light of the degree of impaired frequency resolution as determined by auditory filter shape measurements. [Work supported by NIH.]

Signal Processing Of Aircraft Flyover Noise

A detailed analysis of signal processing concerns for measuring aircraft flyover noise is presented. Development of a de-Dopplerization scheme for both corrected time history and spectral data is discussed along with an analysis of motion effects on measured spectra. A computer code was written to implement the de-Dopplerization scheme. Input to the code is the aircraft position data and the pressure time histories. To facilitate ensemble averaging, a level uniform flyover is considered in the study, but the code can accept more general flight profiles. The effects of spectral smearing and its removal are discussed. Using test data acquired from an XV-15 tilt-rotor flyover, comparisons are made between the measured and corrected spectra. Frequency shifts are accurately accounted for by the de-Dopplerization procedure. It is shown that by correcting for spherical spreading and Doppler amplitude, along with frequency, can give some idea about noise source directivity. The analysis indicated that smearing increases with frequency and is more severe on approach than recession. Simulated spectra were generated using a moving point source model with frequency content similar to the XV-15 data. The results from the simulation corroborated the observations obtained from the analysis of the actual flight data.

Evaluation of a method of simulating reduced frequency selectivity

The accuracy of a method of simulating reduced frequency selectivity by the spectral smearing of complex stimuli has been evaluated. First an excitation pattern that would be evoked by a given nonsmeared stimulus in an impaired ear with broad auditory filters was estimated. Then the spectral smearing of the stimulus that would be necessary to create the same excitation pattern in a normal ear was calculated. The smearing was based on the shapes of simulated broad auditory filters; both symmetric and asymmetric broad filters were simulated. The method was used to process notched noise, and tones in notched noise, and the processed stimuli were used in a series of experiments with normally hearing subjects measuring the threshold for the tone in notched noise. The resulting data were used to derive auditory filter shapes. The derived filter shapes were generally similar to the expected shapes (based on the type of spectral smearing used), but there were some systematic discrepancies and some individual differences. The discrepancies do not seem to be due to the use of information derived from phase locking, since they were observed both at 1 kHz (where phase locking occurs) and at 6 kHz (where phase locking probably does not occurs). The discrepancies also do not seem to be due to the transmission characteristics of the outer/middle ear, since they occurred both when these characteristics were taken into account in the fitting procedure, and when the stimuli were preshaped to compensate for these characteristics. The influence of the subjects' own auditory filters probably can explain some of the discrepancies; the excitation pattern evoked by the spectrally smeared stimuli can be significantly influenced by the subjects' own filters when those filters are not much sharper than the simulated filters used to produce the smeared stimuli. Finally, some of the discrepancies can probably be explained by subjects combining information across auditory filters, rather than just using the single 'best' filter in each condition; this represents a limitation of the fitting procedure rather than of the simulation itself. Overall, the simulation worked reasonably well, especially when the smearing was based on symmetric filters.

Effect of spectral envelope smearing on speech reception. II.

This paper describes two experiments on the effect of reduced spectral contrast on the speech-reception threshold (SRT) for sentences in a background of interfering sound. Signal processing is performed by smoothing the envelope of the squared short-time fast Fourier transform by a convolution with a Gaussian-shaped filter, and overlapping additions to reconstruct a continuous signal. In the first experiment the effect of reduced spectral contrast on the SRT for male speech is investigated and compared with previously obtained results for female speech [ter Keurs et al., J. Acoust. Soc. Am. 91, 2872-2880 (1992)]. Spectral energy is smeared over bandwidths of 1/8, 1/4, 1/3, 1/2, 1, 2, and 4 oct. The results show that, despite the differences in spectral pattern between male and female voices, the SRT in noise increases similarly for both voices for smearing bandwidths over 1/3 oct. In terms of the ripple density of the spectral envelope the results indicate that the range of lower spectral modulations, up to a limit of about 1.5 periods/oct, is sufficient for the intelligibility of speech in interfering sounds. In the second experiment the extent of the threshold difference between a speech masker and a noise masker is investigated for spectral smearing bandwidths of 1/2, 1, and 2 oct. The release from masking found for the speech masker relative to the (steady-state) noise masker decreases with spectral envelope smearing.

Modifications to nonlinear processing for recruitment of loudness.

The time-varying frequency-dependent processing (TVFD) [J. C. Rutledge and M. A. Clements, Proc. IEEE ICASSP, 3641–3644 (1991)] was designed to adjust to changing speech input on both time and frequency basis. Due to inherent problems with the processing technique, some spectral smearing occurs [J. Acoust. Soc. Am. 89, 1974 (A) (1991)]. The processing has been modified to maintain the time-varying frequency-dependent nature, but reduces the smearing effect. A sinusoidal speech model is used with the compensation processing. Appropriate gains for the sinusoid amplitudes are calculated every 5 ms. In the modified processing scheme, the gain determined for the largest peak in each frequency band is applied to all components in that band. This maintains the relative amplitude ratio between neighboring peaks in the frequency spectrum. The compensation algorithm is presented here along with preliminary performance results based on the objective measures described in another paper at this meeting.

Improved Resolution from Retroreflective Screens by Spectral Smearing

AbstractRetroreflective screens are used in many areas of photography. Where it is required to view an image projected in the plane of the screen, the resolution can be limited by the structure of ...

Effects of spectral smearing on speech perception

Connected speech presented in quiet is highly redundant. For such a condition, the resolution of spectral contrasts by the ear seems to be much larger than would be required. This suggests that the ear's high selectivity in frequency is particularly important to understanding speech in the presence of interfering sounds. The effect of reduced frequency resolution on the speech‐reception threshold (SRT) for sentences in noise was investigated for eight normal‐hearing subjects by simulating an auditory system with variable frequency selectivity. Signal processing was performed by short‐time fast Fourier transforms (FFT), reduction of contrast in the spectral envelope without affecting the harmonic structure, and overlapping additions to reconstruct a continuous signal. The Spectral envelope in the frequency region from 100 to 8000 Hz was smeared over fixed relative bandwidths of 1/8, 1/4, 1/2, 1, 2, 4, and 8 octaves. Results show that the SRT increases progressively as the spectral envelope is smeared over bandwidths exceeding the ear's critical bandwidth. In a second experiment phoneme, confusions as a result of three different degrees of spectral smearing are studied in nonsense CVC syllables.

Convective velocities derived from granule contrast profiles in Fe i ? 6569.2

A new determination of the granular and intergranular velocities is described, based on a new approach. The method involves measurement of the granule/intergranule contrast as a function of wavelength on a sequence of filtergrams taken with the CSIRO computer-controlled 1/8 A filter in the photospheric line Fe i λ6569.2. A procedure based on a simple but realistic morphological model of the granulation pattern is used to correct for spatial smearing. The effects of spectral smearing and of scattered light are also taken into account. The present observations reveal a one hundred per cent correlation between brightness and the sense of the vertical velocity component and thus demonstrate beyond doubt the convective origin of the granulation. The new measurements yield a value of ≈1.8 km s−1 for the difference between the upward and downward velocities associated with an average granule. With certain plausible assumptions this leads to granular and intergranular velocities of ≈0.7 km s−1 (upward) and ≈1.1 km s−1 (downward) respectively. Estimates are also obtained for the (true) central intensities and line broadening parameters of the line profile, separately for the average granule and intergranular lane.

Comparison of spectral decomposition methods

Comparison of spectral decomposition methods