Speech Processing 11-492/18-495 Speech Processing 11-492/18-495 - - PowerPoint PPT Presentation

Speech Processing 11-492/18-495 Speech Processing 11-492/18-495

Sound ID

What is in the audio scene What is in the audio scene

 Searching for specific things

Searching for specific things

 Cars, talking, music

Cars, talking, music

 Generically analyzing the audio

Generically analyzing the audio

 Find the “important” parts

Find the “important” parts

Items and Sequences Items and Sequences

 Seq1

Seq1

  Seq2

Seq2

Human judgment is non-standard Human judgment is non-standard

 Context matters

Context matters

 Context includes a priori knowledge not

Context includes a priori knowledge not represented in recording represented in recording

 Two descriptions of a scene from a movie

Two descriptions of a scene from a movie

 “

“There were a series of beeps, and a bomb There were a series of beeps, and a bomb went off” went off”

 “

“A timer counted down, and then there was a A timer counted down, and then there was a big boom” big boom”

4

A Hierarchical Structure for Sound A Hierarchical Structure for Sound

5

Audio data Lower-level units Event sequence Event dependencies

Audio Unit Detection Audio Unit Detection

 Low level acoustic units

Low level acoustic units

 Similar mcep over time

Similar mcep over time

 Find repeated segments over time

Find repeated segments over time

 Find repeated patterns over time

Find repeated patterns over time