Processing Of Complex Sounds Research Articles

Book Review: Processing of Complex Sounds by the Auditory System

Book Review: Processing of Complex Sounds by the Auditory System

Processing of complex sounds in the macaque nonprimary auditory cortex.

Neurons in the superior temporal gyrus of anesthetized rhesus monkeys were exposed to complex acoustic stimuli. Bandpassed noise bursts with defined center frequencies evoked responses that were greatly enhanced over those evoked by pure tones. This finding led to the discovery of at least one new cochleotopic area in the lateral belt of the nonprimary auditory cortex. The best center frequencies of neurons varied along a rostrocaudal axis, and the best bandwidths of the noise bursts varied along a mediolateral axis. When digitized monkey calls were used as stimuli, many neurons showed a preference for some calls over others. Manipulation of the calls' frequency structure and playback of separate components revealed different types of spectral integration. The lateral areas of the monkey auditory cortex appear to be part of a hierarchical sequence in which neurons prefer increasingly complex stimuli and may form an important stage in the preprocessing of communication sounds.

Processing of modulation frequency in the dorsal cochlear nucleus of the guinea pig: Amplitude modulated tones

The modulation frequency (Fm), particularly high Fm (> 200 Hz), in amplitude modulated (AM) tones can elicit the perception of the periodicity pitch (Langner, 1992). In this study, single unit responses to the Fms of the sinusoidal AM tones were investigated at 50 to 90 dB SPL. The recordings were made from the dorsal cochlear nucleus (DCN) of neuroleptic anesthetized guinea pigs with an intact cerebellum. The DCN units show a good capability of phase-locking to Fm at 400–1200 Hz. On-S-type II and Pauser/Buildup (P/B) units have a high modulation gain (7.2–8.3 dB). P/B units can preserve the high modulation gain (5–9 dB) up to 90 dB SPL. The modulation gain exponentially increases with decreasing modulation depth (Dm) and the phase-locking is detectable even at the Dm as low as 2–5%. The ‘central skipping’ of the phase-locking peak has been found at deep Dms in a few cases. The synchronization is independent of the discharge rate and can remain high even when the responses to AM tones are inhibited below the spontaneous activity. Such encoding behaviors over the unit's response area show that the Fm phase-locking is strong near or at its characteristic frequency (CF). The synchronization index (SI) versus carrier frequency (Fc) curve is similar to the inverse shape of tuning curve but more narrowly tuned than the iso-intensity function of pure tones at moderate to high intensity levels. The phase-locking is related to the unit's spontaneous rate (SR). The average modulation gain of the lower SR (≤ 2 spikes/s) units is 5 dB higher than that of the higher SR (> 2 spikes/s) units (8.16 and 2.92 dB, respectively) at 70 dB SPL. These results suggest that AM information is temporally encoded over broad ranges of modulation parameters in the DCN and is conveyed by the Fc channel. Such a timing mechanism can play an important role in processing of complex sounds under normal acoustic conditions.

Auditory-evoked potentials obtained using continuous amplitude-modulated tones

Auditory-evoked potentials (AEPs) recorded from the scalp surface are neurogenic potentials arising from the massed discharge of neural populations in response to acoustic stimuli and provide a ‘‘window’’ on the physiological basis of auditory processing. AEPs, in particular the transient-evoked auditory brain-stem response (ABR) have been widely used in human clinical settings for the evaluation of auditory functioning. Recently, several new techniques using continuous amplitude-modulated tonal stimuli have been investigated. Many of these techniques take advantage of auditory nonlinearities which result in the auditory system acting as an envelope extractor. This envelope following response (EFR), phase locking of neural discharge to waveform periodicities, has been measuring using two-tone, sinusoidal amplitude-modulated, and multitone, multiple envelope components signals. The EFR offers great promise for frequency specific threshold measurements, especially for low-frequency audiometry and for the examination of auditory processing of complex sounds.

Connectionist networks in auditory system modeling

Understanding how complex sounds, such as speech, are processed and eventually perceived in the brain is essential for building more effective speech processors. The echolocating bat provides an animal model for complex-sound processing of identified stimulus features at higher levels of the auditory pathway. In this paper, we present the use of connectionist models for modeling cortical neurons that play a key role in our auditory system model of a species of FM bat, Myotis lucifugus. The influence of network related parameters on modeling accuracy is presented, and the response of these models is explained in a behavioral context.

Book Review: Processing of Complex Sounds by the Auditory System

Book Review: Processing of Complex Sounds by the Auditory System

Population responses to multifrequency sounds in the cat auditory cortex: One- and two-parameter families of sounds

Population responses to multi-frequency sounds were recorded in primary auditory cortex of anesthetized cats. The sounds consisted of single-tone stimuli; two-tone stimuli; and nine-tone stimuli, with the tones evenly spaced on a linear frequency scale. The stimuli were presented through a sealed, calibrated sound delivery system. Single units, cluster activity (CA) and the short-time mean absolute value of the envelope of the neural signal (MABS) were recorded extracellularly from six microelectrodes simultaneously. The CA and MABS were interpreted as measures of the activity of large populations of neurons, in contrast with the single unit activity which is presumably recorded from single neurons. The responses of the MABS signal to simple stimuli were generally similar to those of the CA, but were more stable statistically. Thus, the MABS is better suited for studying the activity of populations of neurons. The responses to tones near the best frequency were strongly influenced by a second tone, even when the second tone was outside the single-tone response area. These influences could be both facilitatory and suppressory. They could not be predicted from the responses to single tones. The responses to the nine-tone stimuli could be explained qualitatively by the responses to the two-tone stimuli. It is concluded that the population responses in primary auditory cortex are shaped by the contributions of the individual frequencies appearing in the stimulus and by the interactions between pairs of frequencies. Interactions between stimulus components are therefore a necessary component of any attempt to explain the processing of complex sounds in the auditory cortex. They may play a role in a global representation of the stimulus spectrum in the primary auditory cortex. The presence of higher-order interactions cannot be excluded by the results presented here.

Philosophy and stimulus design for neuroethology of complex-sound processing.

In research on the neural mechanisms for the processing of biologically important sounds such as species-specific sounds and sounds produced by prey and predators, it is necessary to study responses of central auditory neurons to biologically important sounds, information-bearing elements (IBEs) in them, and tone bursts. The tone bursts or constant-frequency (CF) components can be an IBE in many species of animals. Information-bearing parameters characterizing these sounds must be systematically varied, and tuning of neurons to individual parameters must be studied. The measurement of a tuning curve must be performed not only for excitatory responses, but also for inhibitory and facilitative responses, if any. The selectivity of a neuron to a particular type of sound must be tested for whether it is level-tolerant. Responses to complex sounds can probably be explained on the basis of those to IBEs and tone bursts, so that the use of the tone bursts, even though they are not IBEs, is as essential as that of the biologically important sounds.

Binaural suppression of nonechoes.

A brief diotic conditioner has been shown to effectively disrupt lateralization of a brief dichotic probe presented after a short interval (4-10 ms, onset to onset) with an interaural time delay that is clearly discriminable in the absence of the conditioner [e.g., P. M. Zurek, J. Acoust. Soc. Am. 66, 1750-1757 (1980)], even when the conditioner and the probe are different sounds [P. L. Divenyi and J. Blauert, in Auditory Processing of Complex Sounds, edited by W. A. Yost and C. S. Watson (Erlbaum, Hillsdale, NJ, 1987), pp. 146-155]. The present experiments investigated lateralization suppression of a narrow-band dichotic probe centered at 2 kHz by a narrow-band diotic conditioner, in situations in which the probe could not be regarded as an echo of the conditioner. In one case, the conditioner was identical to the probe but was presented in opposite phase; in other conditions, the center frequency of the conditioner was different, with the center frequency to bandwidth ratio remaining constant. The effects of frequency and temporal separation between the conditioner and the probe on the suppression of the lateralization of the probe were explored. Suppression of lateralization was measured as the increase in the just-noticeable interaural time difference (jnd), with respect to the interaural time jnd obtained for the probe alone. Low-frequency (0.5 to 1.5 kHz) conditioners were more effective in suppressing the lateralization of the probe than those at or above the probe frequency.(ABSTRACT TRUNCATED AT 250 WORDS)

Spectral shape discrimination using three‐tone complexes

Thresholds were obtained for an increment in the intensity of a 100‐Hz, center component of a standard consisting of three equal intensity tones. For three conditions, the frequency separation between each tone was 80, 40, and 10 Hz, respectively. Overall intensity was varied over a 20‐dB range to preclude the use of absolute intensity as a cue. Unexpectedly, thresholds when tones were separated by 80 or 40 Hz differ little from thresholds when tones are separated by more than a critical hand (e.g., profile analysis). Moreover, thresholds increase by 10 dB when the digital‐to‐analog conversion rate is varied randomly between 38–42 μs/sample, providing evidence that listeners use pitch cues. If we assume that listeners use a strategy of off‐frequency listening, then the results are consistent with Feth's EWAIF model [Feth and Stover, in Auditory Processing of Complex Sounds, edited by W. A. Yost and C. S. Watson, pp. 76–86 (1987)]. When tones are separated by 10 Hz, thresholds increase by 5 dB but are unaf...

The detection of a tone in a narrow band of noise

Detection thresholds were obtained for 2000-Hz tones masked by 100-Hz-wide bands of noise centered at 2000 Hz. On each interval of the two-interval, forced choice (21FC) task, the level of the noise or signal-plus-noise waveform was varied over a range of 0 (no variation), 30, or 60 dB. As has been reported previously [Gilkey, in Auditory Processing of Complex Sounds, edited by W. A. Yost and C. S. Watson (Erlbaum, Hillsdale, NJ, 1987); Kidd et al., J. Acoust. Soc. Am. 86, 1310–1317 (1989)], level variation had a negligible effect. Next, thresholds were determined using equal-energy, but uncorrelated, bands of noise. Performance levels in the equal-energy condition were indistinguishable from those obtained using independent noise samples. Thus neither increasing nor decreasing the across-interval variance in stimulus energy altered the detectability of the tone. In a second experiment, equal-energy maskers had bandwidths that varied from 20 to 320 Hz. For this condition, an energy-based ideal detector yields no effect of bandwidth. For bandwidths ranging from 20–80 Hz, human performance was nearly constant. However, thresholds were relatively elevated when the 160- and 320-Hz bandwidths were employed. [Work supported by the National Institutes of Health.]

Auditory Processing of Complex Sounds

Auditory Processing of Complex Sounds

Review of Auditory Processing of Complex Sounds.

Review of Auditory Processing of Complex Sounds.

Echo suppression or localization masking?

A brief dichotic conditioner (C) has been shown to effectively disrupt lateralization of a brief probe (P) presented after a short interval (4–7 ms, onset to onset) with an interaural time delay that is perfectly audible in the absence of the C [P. M. Zurek, J. Acoust. Soc. Am. 66, 1750–1757 (1980)], even when the C and the P are different sounds [P. L. Divenyi and J. Blauert, in Auditory Processing of Complex Sounds, edited by W. A. Yost and C. S. Watson (Erlbaum, Hillsdale, NJ, 1987), pp. 146–155]. An echo suppression mechanism responsible for this effect would predict (1) suppression to be strongest when the C and the P are identical and to decrease monotonically as the spectra of the two sounds are made different, and (2) a monotonic falloff of suppression when the temporal separation between the two sounds is increased beyond a certain minimum. In the present experiments, the effects of frequency separation and temporal separation between a narrow-band P centered at 2 kHz and a C were systematically explored. Contrary to the predictions, low-frequency (0.8 < 1.3 kHz) C's were more effective in suppressing the lateralization of the P than those closer to the P frequency, and lateralization performance was nonmonotonically related to temporal separation between C and P. The results suggest that a dichotic stimulus with a relatively high localization strength could “mask” the localization of another, subsequent dichotic stimulus. [Work supported by the Veterans Administration.]

Perception of harmonic intervals made by simultaneous complex tones

Most studies on pitch perception for complex tones deal with either isolated sounds (pitch matches) or with sequential pairs of such sounds (jnd or melodic interval experiments). Little attention has been given to the question how the auditory system perceives a sound of two or more simultaneous complex tones. Informal observation of musical behavior shows that if a chord, triad, or harmonic interval is played by a group of musical instruments, the various notes usually retain their identity in the perceived sound image. This poses the question how the auditory system parses and groups the spectral components of the total sound in order to compute the correct (simultaneous) pitches. An experiment was performed in which musically trained listeners were asked to identify harmonic intervals made up of two simultaneous complex tones, each comprising three successive random harmonics, by playing the intervals back on a keyboard. Tones were presented both diotically (all six spectral components to both ears) and dichotically (one three-tone complex to each ear). Data show that under most conditions e.g., most of the dichotic ones, the task can be done very well, but that diotic performance depends strongly on details of the total spectral pattern. This suggests that peripheral spectrum and central spectrum impose separate constraints on musical processing of complex sounds.

Correlation between the response selectivity of the auditory thalamic area of the green treefrog and call discrimination

A neurophysiological investigation of the green treefrog, Hyla cinerea, was aimed towards revealing principles underlying call detection. This species was used since its vocal repertoire is not only limited and highly stereotyped, but also controlled behavioral tests have determined the critical parameters that these animals use for species‐specific recognition. In this evoked potential study of the diencephalon, the auditory center was found to be maximally responsive to certain bimodal stimuli. Combinations of two tones with a low‐frequency component around 600 Hz and a high‐frequency component between 2000 and 4000 Hz evoked a potential which was larger than the sum of the potentials to the component tones separately thus revealing a nonlinear convergence underlying the processing of complex sounds. Behavioral studies verify that females respond optimally to playback of synthetic calls having these same bimodal frequencies. Furthermore, temperature changes lead to a shift in the frequencies for the convergent response in the thalamus and for the behavioral selectivity of the females. [Supported by NIH.]

Mechanisms of signal analysis and pattern perception in periodicity pitch.

Progress in the knowledge of auditory processing of complex sounds has been made through coordinated psychophysical, physiological and theoretical studies of periodicity pitch and combination tones. Periodicity pitch is the basis for human perception of musical notes and pitch of voiced speech. The mechanism of perception involves harmonic pattern recognition on the complex Fourier frequency spectra generated by auditory frequency analysis. Combination tones are perceptible distortion tones generated within the cochlea by nonlinear interaction of component stimulus tones. Perception of periodicity pitch is quantitatively accounted for by a two-stage process of frequency analysis subject to random errors and significant nonlinearities, followed by a pattern recognizer that operates very efficiently to measure the period of musical and speech sounds. The basic characteristic of the first stage is a Gaussian standard error function that quantifies the randomness in aural estimation of frequencies of component tones in a complex tone stimulus. Efficient aural measurement of neural spike intervals from the eighth nerve provides a physiological account for the psychophysical characteristic of aural frequency analysis with complex sounds. Although cochlear filtering is an essential stage in auditory frequency analysis, neural time following, rather than details of the filter characteristics, is the decisive factor in determining the precision of aural frequency measurement. It is likely that peripheral auditory coding is similar for sounds in periodicity pitch and in speech perception, although the 'second stage' representing central processing would differ.