Flashcards
speech stuff
Cards in this deck (62)
- 1stops/plosivescomplete vocal tract constriction
burst of energy
papa - 2fricativesnoise with air passing through constriction
shift - 3nasalsinvolve air passing through nasal cavity
e.g mama - 4vowelsopen vocal tract no nasal air
e.g - 5glides/liquidssimilar to vocal tracts but
typically more constriction
wall - 6plosivesbilbo, dada, gaga, pippin, toots, kicks
- 7fricativessea, frank, shfit, this, ship
- 8nasalmama, noon
thing - 9soundis time-variant pressure wave created by vibration
- 10high pressurecompression in the air
- 11lower pressurerarefactions within the air
- 12phase ødisplacement of signal in time
- 13tonesinusoidal function of pressure and
time. - 14amplitudethe degree of displacement in the air.
This is similar to loudness - 15frequencynum of cycles within a unit of time
- 16superpositionadding sinusoids
- 17amp and frequency of the sinusoidsextremely indicative of phoneme
- 18samplingmeasuring the amplitude of
a signal at regular intervals. - 19nyquist ratemin sampling rate necessary to
preserve a signals maximum frequency - 20quantizationfloating point amp -> int
- 21spectograma 3d plot of amp and frequency over time
- 22formantsa concentration of energy within
a frequency band. - 23acoustics of phonemesdepend on context in which phoneme
occours - 24F0 fundamental frequencyrate of vibration of the glottis. Often very indicative of speaker!!!!!!!!!!!!!
- 25sonorantany sustained phoneme in which the glottis vibrates
- 26prosodymodification of speech acoustics in order
to convey extra lexical meaning - 27examples of prosodypitch, duration, loudness
- 28melunit of pitch
- 29mel scale filter bankmimic the response of human ear
often discretize - 30alveolarconstruction by the alveolar ridge /t/
- 31features of phonemesplace, voicing, manner, high/low, front/back
- 32placelocation of primary constriction
- 33voicingglottis is vibrating
- 34mannerclass of articulation
- 35timitphonetic alphabet
- 36affricatesalveolar stop followed by fricative
- 37long tubesproduce deep sounds.
- 38tongue height affectsfirst formant. High tonge = f1 decrease
low tongue = f1 increase - 39tongue frontesscorrelated to second formant, f2
- 40every place of articulationhas voiced and unvoiced vertion.
- 41unvoiced fricativefee, thief, see, she
- 42voicedVendetta, Thee, zha-zha
- 43voiced fricative includesignificant energy
- 44unvoiced plosivesporpoise, tort, kick
- 45voiced plosivesbaboon, dodo, google
- 46voiced stopscharacterized by a voice bar during closure, indicating vibrating glottis
- 47syllablesphonemes grouped together
- 48acoustics of speech caused byglottal pulse waveform passing through vocal tract
- 49important for phoneme recogntiionshape of vocal trac
- 50the cepstrumseparates the source and filter by pretending the log of spectrum is actually a time domain.
spectrum(log(spectrum)) - 51advantages of mfccproduces highly uncorrelated features
- 52GMMweighted linear comb of M components gaussian
- 53Hierarchical clusteringtopdown divisive, agglomerative
- 54divcreates hierarchy by splitting clusters into smaller groups
- 55agglomerativestarts with n and iteratively builds up. two most similar clustersa re merged together
- 56triphone hmmcaptures coarticulation
- 57speaking modeisolated word vs continuous
- 58speaking styleread vs spontanous
- 59R[i-1, j] + 1deletion
- 60R[i-1, j-1]match or perhaps substitution
- 61R[i, j-1] +1insertion
- 62—
Information
- Description
- Flashcards on speech stuff, created by arturo mena on 27/04/2017.
- Created by
- arturo mena
- Views
- 1
- Rating