Edwin D. Burnett

Attack and release times of automatic‐gain‐control hearing aids

The attack and release times were measured on each of three samples of 27 different models of compression‐type hearing aids using the test conditions specified in ANSI Standard S3‐22 (1976). The results indicate that almost all of the hearing aids tested have attack times less than or equal to 10 ms. A little more than half of the hearing aids had release times of 50 ms or less. The range of the attack times varied from 1 to 23 ms and the release times from 5–600 ms.

Polyacusis, a hearing anomaly.

A previously unrecognized anomaly of pitch perception was discovered by H. F. Stimson while testing his hearing. Signals above 3500 Hz did not produce the same pitch in each ear; the right ears sensation was that of a sound at a lower frequency, for which he also had normal pitch perception. Between 3700 and 5200 Hz, the pitch perceived in the right ear was independent of the frequency of the stimulus. As intensity was raised, additional pitches appeared; and for an intense sound, a chord was heard whose constituents were not harmonically related. Stimson can match these anomalous pitches with sinusoidal stimuli of the appropriate lower frequency within about 2%; part of this variation is a change of pitch with intensity. Tones in his normal pitch range having a pronounced overtone structure are perceived as multiple. Overtones falling in the anomalous range give rise to the anomalous pitch. This would indicate a Fourier type of analysis before pitch recognition. Beats are not excited by interference between the pitches evoked by signals in the anomalous range and those heard normally. Loudness studies show signal powers to be additive. Tone‐masking and loudness summation experiments yield data suggesting that the pitch recognition mechanism lies in that part of the auditory system in which loudness is perceived—i.e., beyond the cochlea.

Technique for Varying the Ratio of Harmonic and Intermodulation Distortion in Speech‐Communication Tests

In evaluating the effects of nonlinear distortion upon the intelligibility and naturalness of a transmitted speech signal, it would be desirable to separate intermodulation and harmonic distortion components so that the effects of each could be studied separately. It is not feasible to produce harmonic distortion alone. We have developed a method which produces only difference frequency‐intermodulation components without harmonic distortion. A slight modification produces a signal consisting of harmonic distortion and the sum intermodulation products. The nature of these distorted signals is demonstrated. [This work was supported by the Veterans Administration.]

Pressure and free‐field calibration of “half‐inch” microphones by reciprocity

For measurement situations involving sound fields that are not well specified, uncertainties associated with the choice of diffraction correction to be used for a type L (“one‐inch” nominal diameter) microphone can exceed errors in its calibration by one to two orders of magnitude. One benefit from use of smaller microphones is reduction of these uncertainties. Facilities at the National Bureau of Standards have recently been developed for free‐field and pressure calibration of “half‐inch” nominal diameter laboratory microphones by reciprocity. Both facilities use a grounded shield configuration and insert voltage techniques to determine the open‐circuit sensitivity of a microphone. User‐interactive computer programs provide on‐line computations, display, and storage of calibration results. Adaptors are used to perform pressure calibrations in ANSI S1.10‐1966, (RI971) couplers from 50 to 20 000 Hz; use of other couplers could extend this range. Free‐field calibrations for normal incidence are performed fr...

Some Experiments with Periodically Phase‐Reversed Sinusoids

When periodically phase‐reversed sinusoids are presented to a subject at levels successively raised above auditory threshold, the first perception is of a faint, pure tone. At higher levels a “roughness” is sensed, followed at still higher levels by detectable sidebands. Subjects can match successively the pitches of components of the sidebands; usually, however, just either the sum or difference components are noticed. When reversals occur within just a few periods, the sidebands begin to dominate, but the sensation of the central tone returns if white noise is applied at the contralateral ear. At still shorter intervals between reversals, the central tone can be sensed only at high test levels.

Threshold Perception of Periodically Phase‐Reversed Sinusoids

In connection with a direct study of the integration time of the ear, we generated a test signal by introducing a phase reversal at integral numbers of zero crossings of a sinusoidal signal. When the number of periods between reversal is large, auditory threshold is found at the same level as for a steady‐state pure tone. Subjectively, the phase reversal is sensed as a faint click. In the range 1000–4000 Hz, the threshold is elevated by about 3 dB when the number of periods between reversals is reduced to less than about 50. This is direct evidence of “backward masking.” Ultimately at still shorter intervals between reversals, the sideband components introduced by the interruption are sensed at lower levels than is the central frequency itself, but the central tone is sensed momentarily when an attenuator step is switched, evidently unbalancing the interference.

“Polyacusis” Accompanying Hearing Loss

During audiometric test, H. F. Stimson observed that signals of above 3500 Hz produced different pitches in his right and left ears; the right ears sensation was appropriate to a lower frequency. Upon raising the level, the pitch perceived in the right ear persisted, but another tone appeared. Further investigation of the right ear showed: Above 3700 Hz these tones remain fixed, independent of the frequency of the stimulus that evokes them. Pitch matches can be made either via the left ear, or by additional stimulus at the right ear, but no beats are heard. The tone perceived above threshold is matched by about 2200 Hz. At higher levels, an additional tone corresponding to 1400 Hz appears, and at still higher levels, one corresponding to 1700 Hz. Frequency matches were reproducible within 2% even over weeks. The absence of beats, the tonal purity and constancy of the spurious responses, and their coincidence with pitch sensations evoked by signals of lower frequency imply that the effect originates in a ...

Influence of Distortion Components upon Word Intelligibility

Word lists of the CNC type were recorded on tape via a technique described by Burnett [J. Acoust. Soc. Am. 34, 1997(A), (1962)] that permitted generation of a distorted signal on a second track simultaneous with the linear signal. In Burnetts technique, the signal to be recorded was first passed through a 90° phase splitter and then the distorted signal was obtained by squaring the output. The squaring process applied to the 90° shifted signal produced a phase shift of 180°, so that various of the distortion components could be canceled or enhanced by combining the squared signals from the output of the splitter. By attenuation between the linear and distorted tracks, a number of mixtures of linear and distorted speech could be obtained. The distortion process that enhanced the components most nearly harmonically related to the linear speech signal proved least destructive of intelligibility. The process that enhanced the difference‐frequency intermodulation components was decidedly more efficient in red...

Effects of Harmonic and Intermodulation Distortion on Speech Intelligibility

Although the effects of peak clipping on speech intelligibility have been investigated, the influence of other forms of nonlinear distortion are not as well known. In the present study, we introduced specified amounts of second‐order distortion into PB word lists. A square law circuit generated the distortion products. The playback of the previously recorded PB word lists was split into two channels which fed a dual‐track tape recorder. One channel contained the distortion generator. Thus, while the original clean speech was simply dubbed onto one track, the other track recorded only the second‐order distortion products of that same speech. On playing back the dual‐track tape, any degree of distortion could be obtained by varying the relative amplitudes of the two channels. Intelligibility score (percent of words correct) was investigated as a function of four degrees of distortion, nominally termed 0, 30, 50, and 100% distortion. Findings showed that normals and conductive losses withstood the effects of...

A System for Automatically Recording Third and Higher Order Intermodulation Distortion Components

A problem arises in measuring certain intermodulation distortion components when measurements are made at more than one set of input frequencies. When two input frequencies with a constant frequency difference between them are swept through the frequency range of interest, only some intermodulation components remain fixed in frequency. Others. which may be of significance, particularly in push‐pull systems, change in frequency as the input frequencies are varied. A system is described in which some of these variable frequencies may be tracked by a slightly detuned reference frequency, which is derived by suitable heterodyning of the radio‐frequency signals of two beat frequency audio oscillators. This signal serves as the reference signal in a non‐phase‐sensitive synchronous detector, the output of which is filtered, detected, and automatically recorded. (This work was done under the sponsorship of the Veterans Administration.)