830 likes | 960 Views
Hallucinations in Auditory Perception!!!. Malcolm Slaney Yahoo! Research Stanford CCRMA. Hadoop. One Dimensional (waveform). Pressure. Time. Cochlear Processing. Two Dimensional (not a spectrogram). Cochlear. Place. Time. Correlogram Processing. Three Dimensional (neural movie).
E N D
Hallucinations in Auditory Perception!!! Malcolm Slaney Yahoo! Research Stanford CCRMA
One Dimensional (waveform) Pressure Time Cochlear Processing Two Dimensional (not a spectrogram) Cochlear Place Time Correlogram Processing Three Dimensional (neural movie) Cochlear Place Time Autocorrelation Lag
Correlogram Distance down cochlea Center Frequency Time Interval (s) Autocorrelation Lag With help from Richard O. Duda
Success Reconstructing from correlogram NIPS Keynote
Continuation Tone and Noise Parliament Cough Hear two voices? What do you hear? Waveforms? Ideas? Problems
Pressure Time Cochlear Processing Cochlear Place Time Correlogram Processing Cochlear Place Time Autocorrelation Lag
Speech Examples Wedding Sine Natural
What Vowel is This? Word 1 Word 2 Peter Ladefoged Word 3
Speech Vision Speech Audio Locate Environment Audio Locate Vision Object Speech Speech Object Vision Wedding Vowel? Ventroloquism Dots McGurk Sinewave
ASR Three Three Three Language model for the words: “one”, “two”, “three” Two Two Two One One One Word model showing phonemes for the word one /w/ / / /n/ Acoustic (phoneme) model for the phoneme / / S1 S2 S3
Conventional Scene Analysis Slide by Dan Ellis (Columbia)
Goto—CASA with MIDI MIDI Sequence
Old plus New Principle Slide by Dan Ellis (Columbia)
Saliency Example • Time-frequency display • Saliency map shows high-interest locations
Saliency Maps • Longer tones better • Missing parts salient • Modulation more salient • Forward masking works
Sound Examples • Birds • Calls • Cows • Horse • Waterfall
Saliency Comparison • Details of saliency comparison • Model predictions
X Z M Z M M X Y Y m M Relational Network (Simple) • Patches of neurons • Each measureone quantity • Bidirectionalrelations for feedback/feedforward Thanks to Rodney Douglas
Relational specification Input here Relational feedback RelationalFeedback Relational Network (example)
ASR Relational Network Bidirectional links enforce phoneme/word constraints Phone Recognizer Cochlea Word Recognizer Phone Recognizer Delay A patch of neurons (one of N output) Note: We don’t know how to represent delays
Without A A A I Desired Results Relational Feedback With /A/ Phoneme Patch /I/ Phoneme Patch AI Word Patch IA Word Patch Phoneme Input
ICA Different distributions One Microphone GMM models of distribution Statistical Means
Thanks malcolm@ieee.org
Silicon Frequency Response • Tone ramps into two cochleas
Cochlear Rate Profiles Spikes per utterance Left Cochlea Right Cochlea
Hardware Overview Phoneme Word Cochlea Learning PCI-AER (for remapping) Learning Cochlea Learning Giacomo Indiveri Shih-Chii Liu PCI-AER (for remapping) Implemented in MATLAB