Top Banner

of 39

Cornell Psych 205: day10

May 31, 2018

Download

Documents

James Yu
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
  • 8/14/2019 Cornell Psych 205: day10

    1/39

  • 8/14/2019 Cornell Psych 205: day10

    2/39

    Speech Perception

    demonstrations -speech synthesis & analysiscategorical perception

    discrimination functionsidentification functions

    discriminability = identifiability

    speech production -- source-filter theoryspectrum & spectrogram

    harmonicsformants

    infant & infrahuman perception of speechhabituation & dishabituation

  • 8/14/2019 Cornell Psych 205: day10

    3/39

    speech demos:

  • 8/14/2019 Cornell Psych 205: day10

    4/39

    source

    filter

    Instrumentsto changeshape of filter

  • 8/14/2019 Cornell Psych 205: day10

    5/39

    speech stream:

    vowels: monophthongs: /i/ /a/ /u/, etc.diphthongs: /ei/ /oi/, etc.

    consonants:

    stops: prevoiced: --

    voiced: /b/ /d/ /g/

    voiceless: /p/ /t/ /k/ /?/

    liquids:voiced /l/ /r/ voiceless: --

    fricatives: voiced: /v/ /z/// , etc.voiceless: /f/ /s/// //, etc.

    clicks: -- /!/

  • 8/14/2019 Cornell Psych 205: day10

    6/39

    modified from text

  • 8/14/2019 Cornell Psych 205: day10

    7/39

    harmonic structurefrom source,sawtoothwaveform

    vocal folds/glottis

  • 8/14/2019 Cornell Psych 205: day10

    8/39

    Upper harmonics are integermultiples of the fundamentalfrequency

    schematization of malesource

  • 8/14/2019 Cornell Psych 205: day10

    9/39

  • 8/14/2019 Cornell Psych 205: day10

    10/39

    testosteronetarget inpuberty

  • 8/14/2019 Cornell Psych 205: day10

    11/39

    schematization of femalesource, sawtooth wave of higherfrequency

  • 8/14/2019 Cornell Psych 205: day10

    12/39

  • 8/14/2019 Cornell Psych 205: day10

    13/39

    Chas. Hockett:

    The speech stream is like smashed Eastereggs

  • 8/14/2019 Cornell Psych 205: day10

    14/39

  • 8/14/2019 Cornell Psych 205: day10

    15/39

    demo

  • 8/14/2019 Cornell Psych 205: day10

    16/39

    categorical perception:

    two types of results (for adults)

    identification task

    discrimination task

  • 8/14/2019 Cornell Psych 205: day10

    17/39

    method ofconstant

    stimuli:identification

  • 8/14/2019 Cornell Psych 205: day10

    18/39

    discrimination task:

    stimuli correct response| analog to 2x2cellsresponse stimuli

    1-1 same

    same|same correct rejection

    3-3 samedifferent|same false alarm

    1-3 different

    different|different hit

    3-1 different

    same|different miss

    same for pairs 2-4, 3-5, 4-6, and 5-7

  • 8/14/2019 Cornell Psych 205: day10

    19/39

    same-differentdiscrimination:

    signaldetection task

  • 8/14/2019 Cornell Psych 205: day10

    20/39

    Peak of discriminationfunction at the cross-over point of theidentification functions

  • 8/14/2019 Cornell Psych 205: day10

    21/39

  • 8/14/2019 Cornell Psych 205: day10

    22/39

    baseline

  • 8/14/2019 Cornell Psych 205: day10

    23/39

    /ba/

  • 8/14/2019 Cornell Psych 205: day10

    24/39

    habituation

    to /ba/change in

  • 8/14/2019 Cornell Psych 205: day10

    25/39

    habituation

    habituation to /ba/

    control

    condition #1

  • 8/14/2019 Cornell Psych 205: day10

    26/39

    dishabituationto /da/

    experimentalcondition

  • 8/14/2019 Cornell Psych 205: day10

    27/39

  • 8/14/2019 Cornell Psych 205: day10

    28/39

    lack ofdishabituation to a

    different /ba/

    control

    condition #2

  • 8/14/2019 Cornell Psych 205: day10

    29/39

    infants discriminate /ba/-/da/ like adults do

  • 8/14/2019 Cornell Psych 205: day10

    30/39

    [b][d] [g][k] [r][l]

    adults w/ yes

    adults w/o ---infants yes

    rhesus

    chinchilla

    categorical perception tally

  • 8/14/2019 Cornell Psych 205: day10

    31/39

  • 8/14/2019 Cornell Psych 205: day10

    32/39

  • 8/14/2019 Cornell Psych 205: day10

    33/39

  • 8/14/2019 Cornell Psych 205: day10

    34/39

    [b][d] [g][k] [r][l]

    adults w/ yes yes yes

    adults w/o --- yes noinfants yes yes yes

    rhesus yes ~yes no

    chinchilla ~yes no no

  • 8/14/2019 Cornell Psych 205: day10

    35/39

    Human speech perception takes advantage of what arelikely to have been pre-existing, anisomorphic notches

    in the auditory systems capacity to discriminate

    information.

    Our own evolution may also have created new notches

    Why is this a good thing?

  • 8/14/2019 Cornell Psych 205: day10

    36/39

    speech perception has to be fast.

    speaking rate up to ~300 wpm

    at 4 phonemes/word~20 phonemes/sec

    categorical perception is a way of making rapid

    decisions, throwing away unneeded (withinphoneme category) information

  • 8/14/2019 Cornell Psych 205: day10

    37/39

    lips, jaw, tongue,and pharynx foreating, etc.

  • 8/14/2019 Cornell Psych 205: day10

    38/39

    for speechproduction

    Left

    hemisphere

    for speechandlanguageperception

    Speech Perception

  • 8/14/2019 Cornell Psych 205: day10

    39/39

    Speech Perception

    demonstrations -

    speech synthesis & analysiscategorical perception ---> speed

    & cortical areas

    discrimination functionsidentification functions

    discriminability = identifiability

    speech production -- source-filter theory -->corticalareas

    spectrum & spectrogramharmonicsformants

    infant & infrahuman perception of speech

    h bi i & di h bi i i f t d t