Mel-frequency cepstrum coefficients MFCC are mostly used features in state-of-art speech recognition system O-3 tine 1+ FFT FFT FFT FFT FFT FFT FFTFFT FFTFFT FFT FFTFFT FFT Spectrum Cepstral Vectors
Mel-frequency cepstrum coefficients MFCC are mostly used features in state-of-art speech recognition system 1/28/2021 HUMAN COMPUTER INTERACTION 11
Mel-frequency cepstrum coefficients Speech waveform 05 0.5 15 2.5 Log(mel) filterbank energies Mel frequency cepstrum 05 15 25 HUMAN COMPUTER INTERACTION
Mel-frequency cepstrum coefficients 1/28/2021 HUMAN COMPUTER INTERACTION 12
Statistical speech recognition Recorded speech D coded lext (Transcription Hidden Markov Model SIg Analysis A co ustIc Model Search axIcon Tr ace rainin Data anguage yodel n-gram model HUMAN COMPUTER INTERACTION
Statistical speech recognition 1/28/2021 HUMAN COMPUTER INTERACTION 13
Outline Introduction Speech recognition based on HMM ● Acoustic processing Acoustic modeling: Hidden Markov Model anguage modeling HUMAN COMPUTER INTERACTION
Outline Introduction Speech recognition based on HMM • Acoustic processing • Acoustic modeling: Hidden Markov Model • Language modeling 1/28/2021 HUMAN COMPUTER INTERACTION 14
Acoustic modeling: Hidden Markov Model Acoustic models, P(OW), are used to compute the probability of observing the acoustic evidence o when the speaker utters The variability in the speech signal due to factors like environment, pronunciation, phonetic context, physiological characteristics of the speaker make the estimation a very complex task The most effective acoustic modeling is based on a structure referred to as Hidden Markov Models(HMm HUMAN COMPUTER INTERACTION
Acoustic modeling: Hidden Markov Model Acoustic models, 𝑃(𝑂|𝑊), are used to compute the probability of observing the acoustic evidence O when the speaker utters W. The variability in the speech signal due to factors like environment, pronunciation, phonetic context, physiological characteristics of the speaker make the estimation a very complex task. The most effective acoustic modeling is based on a structure referred to as Hidden Markov Models (HMM). 1/28/2021 HUMAN COMPUTER INTERACTION 15