Term
| 4 ways of measuring speech production |
|
Definition
| 1. Stimulus 2. Task 3. Measure 4. Population |
|
|
Term
| Speech Production Stimuli (3) |
|
Definition
1. Natural Speech--ex: can chinese tell the difference between English vowels?
2. Modified Natural Speech-how vowel duration affects perception
3. Synthetic Speech-how format frequencies affect perception. |
|
|
Term
|
Definition
| Identification, Discrimination, Matching |
|
|
Term
| Speech Perception Measure |
|
Definition
-Response Choice (look at percievers actual responses)
-Response Time (how long it takes perciever to respond) |
|
|
Term
| Speech Perception Population |
|
Definition
| -Clinical Population, Developmental Status, Speaker's Language, Listener's Language |
|
|
Term
| How do we perceive speech? |
|
Definition
Traveling waves on the cochlea are activated at frequency dependent locations.
-Basilar Membrane does a Fourier transform, decomposing speech into its constituent waves. |
|
|
Term
| A traveling wave will show peaks on what? |
|
Definition
| -Basilar Membrane at locations determined by formants of speech |
|
|
Term
|
Definition
| Peaks of spectral energy that arise from filtering of the vocal source |
|
|
Term
|
Definition
| Rapid changes in formant frequencies as articulation moves from consonant to vowel or visa versa |
|
|
Term
| Coarticulation Definition |
|
Definition
| phonemes arent produced in isolation, but rather overlap in time. Articulation of a phoneme depends on what came before and after. |
|
|
Term
| Due to overlap of phoenemes leaving no space in Coarticulation..what 2 problems occur? |
|
Definition
1. Lack of Segmentability
2. Lack of invariance |
|
|
Term
|
Definition
-speech stream cannot be broken down to discrete segments
-any segment contains info about itself and sounds surrounding it |
|
|
Term
|
Definition
| No necessary/sufficient attribute uniquely specifies a given speech sound |
|
|
Term
| Lack of Segmentability and Invariance together creates what general problem? |
|
Definition
| -the same acoustic info can signal different sounds |
|
|
Term
|
Definition
| by allowing air to pass unimpeded through the vocal tract |
|
|
Term
| Why do "Different" vowels affect formants? |
|
Definition
| bc we move our tongue back/forth and up/down |
|
|
Term
| how do we identify different vowels? |
|
Definition
|
|
Term
| According to the source/filter theory what determines formant frequency? |
|
Definition
| length of the vocal tract |
|
|
Term
| larger people have longer vocal tracts therefore... |
|
Definition
| they have lower formants! |
|
|
Term
| What 2 types of info is a speaker conveying when the produce utterances? |
|
Definition
| Linguistic properties & Indexical properties |
|
|
Term
| linguistic properties of utterances carry what...? |
|
Definition
| speakers intended message |
|
|
Term
| What 2 types of info is a speaker conveying when the produce utterances? |
|
Definition
|
|
Term
| Indexical properties of utterances carry what.. |
|
Definition
| info about the speakers voice |
|
|
Term
| The way that any speech sound is produced depends on what 2 things? |
|
Definition
| Phonetic Variables & Speakers Variables |
|
|
Term
| What is the process of Talker Normalization |
|
Definition
| stripping away the indexical properties of an utterance to get at the linguistic properties. |
|
|
Term
| In vowel normalization the absolute values of formant frequencies reveal what? |
|
Definition
| Information about the Speaker (speaker variables) |
|
|
Term
| In vocal normalization the relative values of formant frequencies reveal what? |
|
Definition
| Info about vowel identity (linguistic variables) |
|
|
Term
| Proposed Transformations that allows a listener to calibrate a speaker's vowel space |
|
Definition
-logarithms of formant frequencies
-ratios of formant frequencies
-logarithms of ratios of formants frequencies |
|
|
Term
| Pro of Ratios of Logarithms |
|
Definition
| -we can reduce or eliminate differences due to talkers age and sex |
|
|
Term
| Con of Ratios of Logarithms |
|
Definition
| -still variability--different vowels can have the same ratios ex: see /u/ /æ/ & /ʊ/ |
|
|
Term
| How does listener "calibrate" speakers vowel space? |
|
Definition
| Schwa, Point Vowels, Any Vowel |
|
|
Term
| NOrmalizing via Schwa ( /ə/ ) |
|
Definition
| mid-central vowel possibly used to calibrate speakers vowel space |
|
|
Term
| Normalizing via Point Vowels |
|
Definition
Maybe listeners use the point vowels, /i/, /u/, and /a/
-they are extreme vocal tract positions and formant freqeuncy values |
|
|
Term
| Normalizing via Any Vowel |
|
Definition
Listeners are just as good with central vowels as with point vowels.
/i/, /a/, and /u/
/ɪ/, /a/, and /ʌ/ |
|
|
Term
| why is vowel perception easier than consonant perception? |
|
Definition
| Vowels tend to be voiced, have high amplitude, clearer formants, and longer durations |
|
|
Term
| motor theories indicate.. |
|
Definition
that speech is perceived in terms of how we produce it.
-->percieving gestures rather than phonemes
articulation>sounds |
|
|
Term
| Who came up with the motor theory |
|
Definition
|
|
Term
| Who came up with Direct Realism? |
|
Definition
|
|
Term
| Motor Theories 3 main ideas about perception |
|
Definition
1. Perception is based on production
2. Perception is species-specific
3. perception is innate |
|
|
