Week7 - Audition Vision: use eyes/camera and extract...

Info iconThis preview shows pages 1–5. Sign up to view the full content.

View Full Document Right Arrow Icon
Audition ! Vision: use eyes/camera and extract perceptual information ! Audition: use ears/microphone and extract perceptual information ! A little experiment… Audition tasks ! Computational Auditory Scene Analysis (CASA) ! Separate out different sound signals ! Speech enhancement ! Similar to CASA, but focused on extracting speech from other signals (noise, other speech) ! Music analysis ! Analysis of musical styles ! Music information retreival Pitch ! Many of these applications depend on pitch ! Simplest definition: ! Sounds are vibrations of air molecules at a particular frequency frequency = 1/wavelength 1 Hertz (1 Hz) = 1 vibration/sec ! Often, vibrations occur at multiples of the fundamental frequency Strength of overtones gives instruments different sounds, for example ! Pitch is (to a first approximation) the fundamental frequency Fundamental frequency @ 200 Hz Adding in a harmonic @ 400Hz Complex signal: Summing the two harmonics
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
A yet more complex signal: 4 harmonics Why pitch is perceptual rather than analytical ! Effect of removing harmonics on pitch ! Strike note of chime (4:29) ! Noise as pitch (6:54) ! Circularity of pitch How humans do it ! The ear has several parts that allow it to pick up on acoustic frequencies ! Eardrum: transmits sounds from outer ear to 3 bones in middle ear (ossicles) ! Ossicles transmit sound vibrations to cochlea ! Cochlea: a fluid-filled snail-shaped tube Hair cells: act as frequency triggers inside cochlea Cochlea is organized tonotopically (in frequency order): high frequency near ossicles, low frequency further down basilar membrane A cross-section of the cochlea source: Wikipedia Machine approximations to the cochlea ! Filterbank: a bunch of overlapping frequency filters centered at various points along the frequency spectrum f Filterbank response to 200 Hz Pulse
Background image of page 2
200 Hz, different filterbank 250 Hz, same filterbank 333 Hz, same filterbank Autocorrelation ! You can (usually) find the fundamental frequency by comparing a signal to shifted versions of itself (Autocorrelation) ! R(j)= ! x n x n-j Autocorrelation ! You can (usually) find the fundamental frequency by comparing a signal to shifted versions of itself (Autocorrelation) ! R(j)= ! x n x n-j Autocorrelation ! You can (usually) find the fundamental frequency by comparing a signal to shifted versions of itself (Autocorrelation) ! R(j)= ! x n x n-j
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Autocorrelation ! You can (usually) find the fundamental frequency by comparing a signal to shifted versions of itself (Autocorrelation) ! R(j)= ! x n x n-j Autocorrelation ! You can (usually) find the fundamental frequency by comparing a signal to shifted versions of itself (Autocorrelation) ! R(j)= ! x n x n-j optimal lag Mid-level representations form the basis for segment formation and subsequent grouping processes Correlogram extracts periodicity information from frequency analysis Summary correlogram can be used to identify F0 Cross-correlation between adjacent correlogram channels
Background image of page 4
Image of page 5
This is the end of the preview. Sign up to access the rest of the document.

Page1 / 12

Week7 - Audition Vision: use eyes/camera and extract...

This preview shows document pages 1 - 5. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online