This preview shows page 1. Sign up to view the full content.
Unformatted text preview: CSD 250 CSD 250 Speech Science
Speech perception The Speech chain The Speech chain Speech perception Speech perception R1 R0 [p]
100 80 Amplitude (dB) 60 40 20 0 0 2 4 6 Frequency (kHz) 8 10 [� ] [p] “pipe” 250 230 Frequency (Hz) 210 190 170 150 0 105 210 Time (ms) 315 420 525 Challenges in Challenges in speech perception Segmentation problem Segmentation problem He is a good baby. Where is sound boundary? Where is sound boundary? I owe you a yoyo. Contextual variability: Contextual variability: Lack of acoustic invariance Speaker variability Speaker variability Hearing in speech Hearing in speech perception Hearing is no doubt necessary for speech perception. But is it sufficient to account for speech perception? Example 1: Phoneme restoration Example 2: McGurk effect Example 2: How is speech perceived? How is speech perceived? Acoustic signal is continuous Linguistic representation appears to be discrete or categorical (e.g. phonemes, distinctive features) How is the mapping achieved? How does a baby learn about the association? Acoustic cues Acoustic cues Not everything in acoustic signal is critical for speech perception (e.g. word pitch, voice quality) Only certain aspects seem to be relevant for perceiving phonetic dimensions of speech Synthetic speech is used to evaluate acoustic cues Acoustic cues for vowels Acoustic cues for vowels heed hid head had hod hawd hood Who’d First 2 (or 3) formants First 2 (or 3) formants Speaker variability? Speaker variability? Vowel normalization Intrinsic cues: F0 & higher formants within a vowel Extrinsic cues: phonetic context (i.e. other vowels) Consonant perception: Consonant perception: Stops as example Manner: burst, short formant transition Place: burst freq, formant transition Voicing: F1 cutback, VOT, vowel length Multiple acoustic cues Multiple acoustic cues Some cues for stop voicing
– “Voice bar”: lowfrequency periodicity – Voice onset time (VOT) – Preceding vowel duration: for syllable final obstruents Voice bar bad pad boot Voice onset time Voice onset time Vowel duration as cue to Vowel duration as cue to stop voicing bad bat cab cap Stop place perception: Stop Formant transition Categorical perception Categorical perception Another example of CP: Another example of CP: Voicing Why is CP significant? Why is CP significant? CP appears to be specific to speech but not to nonspeech analogs Why is CP significant? Why is CP significant? Infants as young as 1 and 4month old demonstrate CP for stop voicing Speech is special? Speech is special? Infants: Biologically endowed for language? A “speech mode” distinct from general auditory process? The Motor Theory “Encoded” nature of speech
– – – (Seemingly) lack of acoustic invariance Multiple acoustic cues Articulatory gestures as target of perception Not so fast: We show CP too Not so fast: We show CP too ...
View Full Document
- Winter '10