M. Jane Collins

Discrimination of formant transition onset frequency: Psychoacoustic cues at short, moderate, and long durations

Two experiments determined the just noticeable difference (jnd) in onset frequency for speech formant transitions followed by a 1800-Hz steady state. Influences of transition duration (30, 45, 60, and 120 ms), transition-onset region (above or below 1800 Hz), and the rate of transition were examined. An overall improvement in discrimination with duration was observed suggesting better frequency resolution and, consequently, better use of pitch/timbre cues with longer transitions. In addition, falling transitions (with onsets above 1800 Hz) were better discriminated than rising, and changing onset to produce increments in transition rate-of-change in frequency yielded smaller jnds than changing onset to produce decrements. The shortest transitions displayed additional rate-related effects. This last observation may be due to differences in the degree of dispersion of activity in the cochlea when high-rate transitions are effectively treated as non-time-varying, wideband events. The other results may reflect mechanisms that extract the temporal envelopes of signals: Envelope slope and magnitude differences are proposed to provide discriminative cues that supplement or supplant weaker spectrally based pitch/timbre cues for transitions in the short-to-moderate duration range. It is speculated that these cues may also support some speech perceptual decisions.

Audibility of short-duration tone-glides as a function of rate of frequency change

Thresholds for rising and falling tone-glides were determined against a background of 50-6000 Hz noise at a level of 60 dB re 20 microPa. Glides were centered around 2000 Hz and changed frequencies linearly at rates of 24, 48, 96 and 192 Hz/ms; tone-glide durations were 5, 10, 20 and 40 ms. Results demonstrate a rate-dependent asymmetry in the detectability of rising and falling tone-glides, with rising tone-glides detected at lower signal intensities for the higher rates of frequency change (i.e., 96 and 192 Hz/ms).

The time course of acoustic/phonemic cue integration in the sensorineurally hearing‐impaired listener

There is limited documentation available on how sensorineurally hearing-impaired listeners use the various sources of phonemic information that are known to be distributed across time in the speech waveform. In this investigation, a group of normally hearing listeners and a group of sensorineurally hearing-impaired listeners (with and without the benefit of amplification) identified various consonant and vowel productions that had been systematically varied in duration. The consonants (presented in a /haCa/ environment) and the vowels (presented in a /bVd/ environment) were truncated in steps to eliminate various segments from the end of the stimulus. The results indicated that normally hearing listeners could extract more phonemic information, especially cues to consonant place, from the earlier occurring portions of the stimulus waveforms than could the hearing-impaired listeners. The use of amplification partially decreased the performance differences between the normally hearing listeners and the unaided hearing-impaired listeners. The results are relevant to current models of normal speech perception that emphasize the need for the listener to make phonemic identifications as quickly as possible.

Effects of background noise level on detection of tone glides

Detection thresholds were obtained for short duration rising tone glides and falling tone glides for normal hearing listeners in quiet and with varied levels of background, broadband noise. For signal durations 50 ms, or greater, pure tones were detectable at lower levels than were rising and falling tone glides; no level effects were observed. For signal durations less than 50 ms an interaction between background noise level, signal duration, and signal class was observed. The previously reported tendency for short duration rising glides to be detected at lower sound pressures than pure tones or falling tone glides appears to be unique to a limited range of noise levels (40-75 dB).

Temporal integration of two‐component tone glides

Temporal integration of complex signals composed of two linearly combined tone glides was studied in a 60-dB SPL noise background. When both glide tones increased or decreased in frequency, integration functions differed for durations shorter than 50 ms in a manner similar to results obtained for single-component tone glides. That is, rising two-component glides were detected at lower thresholds than were falling two-component glides. However, when one component increased in frequency and the other decreased, thresholds were the same, regardless of which component increased and which decreased.

Speech production skills of children with cochlear implants with congenital or acquired deafness

Earlier results [J. Acoust. Soc. Am. 88, S192 (1990)] showed that children with early acquired deafness demonstrated speech production skills superior to those of children with congenital deafness both before and after they had received cochlear implants. In the previous study, however, speech production changes were analyzed without regard for type of implant used by the child. This study examined changes in speech production in terms of age at onset of deafness (congenital or acquired before age 3) and type of implant (single‐or multichannel). Speech samples were obtained in the pre‐implant condition and at four, 6‐month post‐implant intervals. The subjects with acquired deafness produced phonetic features with greater accuracy than did the subjects with congenital deafness in the pre‐implant condition. At all post‐implant intervals, the subjects with acquired deafness who received the single‐channel implant demonstrated better speech skills than those of the children with congenital deafness. In contra...

A procedure for the study of auditory stream segregation

An approach based on signal detection theory was used to measure the stream segregation of tone sequences. Previous research [Dannenbring and Bregman, J. Exp. Psych. Human Percept. Perform. 2, 544–555 (1976)] has shown that stream segregation is highly correlated with the elements in a sequence being perceived as ‘‘overlapped.’’ In the present experiment, two classes of sequences were employed. One half of the sequences was composed of tones that physically overlapped, and the remainder of the sequences in the experiment used tones which were physically separated. The sequences were parametrically varied in terms of duration of the sequences and frequency differences of the tones in the sequences. The experimental task required the subject to decide whether or not a sequence was overlapped. Responses to the two classes of sequences were used to calculate measures of discriminability, bias, and criterion. Results using this procedure primarily show significant main effects of frequency difference of the el...

Spectral shaping options for persons with low‐ or high‐frequency sensorineural hearing loss

When hearing loss is limited to one particular frequency region, amplification is traditionally provided in the region of the poorest audiometric thresholds. To the extent that audiometric data correlate with auditory system damage, the usefulness of information provided to these regions of greatest damage is questionable. Optionally, gain may be provided to frequency regions where thresholds are improving towards normal. To that end, continuous discourse intelligibility ratings and consonant discrimination scores were obtained from individuals with either low‐frequency or precipitously sloping high‐frequency sensorineural hearing loss. During testing, spectral emphasis was variously placed in regions of hearing impairment and/or improving thresholds. For the low‐frequency impaired subjects, consistently inferior subjective ratings and discrimination scores were obtained when spectral emphasis was placed solely in the region of hearing loss. For the high‐frequency impaired subjects, no single spectral wei...

Second formant transition discrimination

The ability of listeners to process information conveyed by second formant transitions may be expected to be limited by their ability to detect differences in the rate of frequency change for the transition. To examine this question, difference limens (DLs) for initial transitions of isolated second formants were established using a four interval “same‐different” procedure. These single formant signals were 300 ms in duration (60‐ms transitions and 240‐ms steady‐state portions). The DLs for starting frequency, and thus rate of frequency change, were established for three standards which varied in rate and direction of transition; all had steady states at 1800 Hz. One standard had no transition and two had transitions of 2.5 Hz/ms that either fell from 1950 Hz or rose from 1650 Hz. The DLs were established for transitions with starting frequencies above and below each standard condition. The DLs for three of four experienced listeners were smaller for transitions in the frequency range above the steady state than for transitions in the frequency range below the steady state. Mean discriminable differences in starting frequency were 67 and 87.5 Hz in the frequency ranges above and below the steady state, respectively. [Supported by NINCDS and the Louisiana Lions Foundation.]The ability of listeners to process information conveyed by second formant transitions may be expected to be limited by their ability to detect differences in the rate of frequency change for the transition. To examine this question, difference limens (DLs) for initial transitions of isolated second formants were established using a four interval “same‐different” procedure. These single formant signals were 300 ms in duration (60‐ms transitions and 240‐ms steady‐state portions). The DLs for starting frequency, and thus rate of frequency change, were established for three standards which varied in rate and direction of transition; all had steady states at 1800 Hz. One standard had no transition and two had transitions of 2.5 Hz/ms that either fell from 1950 Hz or rose from 1650 Hz. The DLs were established for transitions with starting frequencies above and below each standard condition. The DLs for three of four experienced listeners were smaller for transitions in the frequency range above the steady sta...

Measurement of the 2f1‐f2 combination tone in a subject with a noise‐induced hearing loss

A cancellation technique was used to measure the 2f1‐f2 combination tone in a subject with a bilateral asymmetric noise‐induced hearing loss. The study was designed to determine the effects of primary frequency and level on the perception of the 2f1‐f2 combination tone. The f2/f1 ratio was 1.2, and six primary tone pairs were chosen such that their frequencies fell below, in, and above the frequency region of the notched hearing loss. In the subjects poorer ear, the 2f1‐f2 combination tone was not perceived when the combination tone frequency fell in the region of the hearing loss. When the primary frequencies fell in the region of the hearing loss, the 2f1‐f2 combination tone was not perceived for input levels of 50 and 60 dB SPL but was for 70 dB SPL.