Summary Discussion on Computational Audition

1
Summary Discussion on Computational Audition
Led by DeLiang Wang
2
A Comparison between Vision and Audition

• Visual features
• Color/luminance
• Orientation
• Location
• Depth
• Motion
• Texture
• Size and shape

• Auditory features
• Frequency, spectrum
• Pitch
• Location
• Onset/offset
• Motion (Doppler effect)
• Texture?
• AM/FM
• Rhythm, duration

3
A Feature Integration Theory for ASA?

• Adapted from Treisman86

4
10 Big Problems in Computational Audition?

• Sound localization and location-based source
  segregation
• With two sensors
• Rapid adaptation to acoustic environment
• Echo and reverberation
• Sound motion analysis (Demo )
• Sound texture analysis ( )
• Event tracking/prediction
• Rhythm
• Sequential pattern analysis
• Role of attention

• Modeling and classification of natural sound
  sources
• Emphasis on non-speech, non-music environmental
  sounds
• Target detection in auditory scenes
• Monaural segregation of unvoiced speech
• Multicue integration for auditory scene analysis
• Double-integration problem
• Neural substrate of auditory scene analysis

5
Computational Audition

• A field of study, like computational vision?