Search

The effect of duration on the visual‐speech intelligibility of talkers was investigated. The stimulus set consisted of 25 sentences. Each sentence had the same grammatical structure and contained three critical elements (subject, verb, object). Seven talkers were videotaped while they spoke a list of sentences twice using conversational and clear speech. For each talker, the mean duration of the conversational and clear sentences were measured. The recordings were digitized and dubbed under four conditions: (1) normal conversational; (2) conversational speech decelerated to a speed equivalent to the duration of the talker’s average utterances of clear speech; (3) natural clear speech; (4) clear speech accelerated to a speed equivalent to the talker’s average conversational speech. The test sentences (25 sentences × 2 iterations × 4 durations × 7 talkers) were randomized and shown (without sound) to a group of 18 subjects. The responses obtained during the perceptual task were used to determine each talker’s speech intelligibility for the four experimental conditions. The results revealed a significant effect for sentence duration and talker, as well as a significant sentence‐duration x talker interaction. The effects of sentence duration on visual‐speech intelligibility will be discussed. [Work supported by NSERC.]

Iterations of sentences were recorded audio‐visually from talkers while they participated in a speech‐tracking task. Six female talkers produced iterations of conversational and clear speech under two different experimental conditions: (a) while the talker was informed that only her visual–speech cues would be transmitted to the interlocutor and (b) while she was informed that only her auditory–speech cues would be transmitted to the interlocuter. In reality, both her auditory– and her visual–speech cues were recorded under each experimental condition. Target sentences were extracted for the recordings, edited, and presented in a random order to a group of 48 subjects. The subjects completed a speech‐recognition task under two perceptual modalities: auditory‐only and visual‐only. The subjects’ mean speech‐recognition scores were used to determine the speech intelligibility scores of individual talkers for each experimental condition. The results failed to reveal any differences between the speech intelligibility scores obtained while a talker intended to produce iterations of visual‐clear speech and those obtained while she intended to produce iterations of auditory‐clear speech. Hence, the findings failed to demonstrate that talkers modify their articulation patterns in order to compensate for the perceptual modality under which the interlocutor receives the speech information. [Work supported by a NSERC grant awarded to J‐PG.]

The visual speech intelligibility of /C–v/ and /C–v–C/ syllables spoken under conversational‐like and clear speech conditions was compared. The syllables consisted of voiced consonants, varying in place of articulation (i.e., /b,d,g,v,z,j/), presented in three vowel contexts (/a,i,u/). Talkers were six female adults, and each produced four iterations of the stimulus set for each of the two speaking conditions. The 1728 videotaped items (36 syllables ×2 manners of speech ×4 iterations ×6 talkers) were edited, randomized, and shown (without sound) to viewers with normal hearing. Viewers’ completed a consonant‐recognition task and these data were used to determine a visual speech‐intelligibility score for each stimulus set produced under the conversational and clear speech conditions. Results revealed that, despite significant within‐talker variability in both speaking conditions, significantly higher speech intelligibility scores were observed for the production of clear speech. A significant talker effect,...

Three experiments were conducted to determine whether listeners with a sensorineural hearing loss exhibited greater than normal amounts of masking at frequencies above the frequency of the masker. Excess masking was defined as the difference (in dB) between the masked thresholds actually obtained from a hearing-impaired listener and the expected thresholds calculated for the same individual. The expected thresholds were the power sum of the listener's thresholds in quiet and the average masked thresholds obtained from a group of normal-hearing subjects at the test frequency. Hearing-impaired listeners, with thresholds in quiet ranging from approximately 35-70 dB SPL (at test frequencies between 500-3000 Hz), displayed approximately 12-15 dB of maximum excess masking. The maximum amount of excess masking occurred in the region where the threshold in quiet of the hearing-impaired listener and the average normal masked threshold were equal. These findings indicate that listeners with a sensorineural hearing loss display one form of reduced frequency selectivity (i.e., abnormal upward spread of masking) even when their thresholds in quiet are taken into account.

The long‐term average speech spectrum (LTASS) was obtained from ten male and ten female talkers for two microphone positions. Dual channel recordings of speech were recorded and played back for spectrum analysis. One channel recorded the speech signal obtained from a microphone placed 30 cm directly in front of the talker, while the second channel recorded the speech signal obtained from a microphone placed at the talker's ear. A spectrum analyzer (BK type 4132) was used to compute the overall levels and 1/3‐octave‐band levels. The analyses indicated significant differences in overall and the 1/3‐octave‐band levels between the two recording positions. The mean overall level at the position of the ear was approximately 6 dB higher than the overall level at the standard recording position. Also, 1/3‐octave‐band analyses indicated that the ear‐level LTASS consisted of more low‐frequency energy and less high‐frequency energy than the LTASS recorded at the standard position. The implications of these results for hearing aid selection will be addressed.

The effects of source function, frequency, and bandwidth on the discriminability of resonant frequencies of synthetic one‐formant stimuli were investigated. Two periodic signals approximating glottal waveforms (F0 = 100 Hz; F0 = 250 Hz) and broadband noise were used to excite a voltage‐controlled filter whose transfer function resembled that of a formant. The periodic waveforms were generated with either no change in F0, or with a 50‐Hz rise and fall in F0. Four resonant frequencies (fr = 300, 475, 650, 800 Hz) and four stimulus bandwidths (Q = 1.0, 2.8, 10.0, 36.0) were investigated. In general, the results show no systematic differences among the DL's obtained with periodic waveforms and the noise source. Also, as expected, the results show that the DL's increase as the fr increases and decreases with increasing Q. For the experimental conditions investigated, the results are well summarized by the following equation: Δfr = 0.079 fr/√Q.