Auditory-phonetic projection and lexical structure in the recognition of sine-wave words.

Abstract

Speech remains intelligible despite the elimination of canonical acoustic correlates of phonemes from the spectrum. Listeners tolerate distortion or spectral blur in tone analogs, noise band vocoded speech, and acoustic chimeras in utterances ranging from syllables to isolated words and sentences. A portion of this flexibility is attributable to short-term perceptual learning in auditory-to-phonetic projection, though exploiting the properties of lexical neighborhoods plays a role with utterances composed of words. New tests were conducted to estimate talker learning, segmental sensitivity, and lexical knowledge in this kind of perceptual versatility. Sine-wave versions of the easy/hard word lists were created, and the performance-level difference between the two lists was used to index the default reliance on lexical processes. Several kinds of preliminary exposure were used to induce sensitivity to a dimension of perceptual learning: sine-wave speech produced by the same talker, sine-wave speech of a different talker, natural speech of the same talker. A comparison of exposure effects on the performance level of easy/hard word recognition offered a clue about the differential contribution of talker-based, segment-based, and lexically-based attention in speech perception without canonical spectra. Implications for perceptual accounts based on cue likelihood will be discussed. [Work supported by NIDCD.]

A neural mechanism for recognizing speech spoken by different speakers

“…The behavioral results of the present study are consistent with previous findings showing that speech recognition is influenced by the acoustic characteristics of speakers. A number of behavioral findings suggested that attributes of a speaker's voice are perceived and memorized along with the speech message (Bradlow et al, 1999;Palmeri et al, 1993;Pisoni, 1993) and that knowledge about the vocal characteristics of a speaker enhances comprehension of what is said (Best et al, 2008;Bradlow and Pisoni, 1999;Kitterick et al, 2010;Levi et al, 2011;Nygaard and Pisoni, 1998;Nygaard et al, 1994;Remez et al, 2009;Yonan and Sommers, 2000). Conversely, understanding speech in the context of speaker changes typically results in performance costs and longer processing times (e.g., Magnuson and Nusbaum, 2007).…”

Infants Exposed to Fluent Natural Speech Succeed at Cross-Gender Word Recognition

“…Error bars represent standard errors of the difference scores. Remez et al, 2011). Note, however, that under everyday listening conditions, infants typically receive a large proportion of their language input from speakers who spend a lot of time together and thus may exhibit this type of vocal convergence.…”

Intelligibility improves perception of timing changes in speech

“…This was done using Praat software (version 6.12, from http://www.fon.hum.uva.nl/praat/download_win.html) and a script written by Chris Darwin (http://www.lifesci.sussex.ac.uk/home/Chris_Darwin/Praatscripts/SWS). While it is well established that the intelligibility of algorithmically-constructed sine-wave speech is inferior to that constructed by hand 40 , for our purposes this script was sufficient to create a stimulus that is intelligible to informed listeners, but not to naïve listeners. The resulting sine-wave speech sentences (Fig.…”

“…4A) for this purpose, a stimulus created from sets of three sine waves that track the amplitude and frequency of speech formants 39 . Sine-wave speech is of particular interest for this purpose as its intelligibility can be very quickly and dramatically affected by perceptual learning: Naïve listeners, who have never been exposed to sine-wave speech, typically perceive the stimulus as non-speech sounds, such as whistles or bird songs, whereas listeners trained to perceive it as speech can often identify the different words in the stimulus 40 . This property of sine-wave speech has been used previously in, for example, brain imaging experiments to compare brain responses to identical stimuli that are heard as speech or non-speech [42][43][44] .…”