A Multimode 157μW 4-Channel 80dBA-SNDR Speech-Recognition Frontend With Self-DOA Correction Adaptive Beamformer.
Taewook KangSeungjong LeeSeungheun SongMohammad R. HaghighatMichael P. FlynnPublished in: ISSCC (2022)
Keyphrases
- speech recognition
- direction of arrival
- hidden markov models
- speech synthesis
- speech processing
- pattern recognition
- minimum variance
- language model
- automatic speech recognition
- speech signal
- speech recognizer
- doa estimation
- noisy environments
- speech understanding
- speaker identification
- speech recognition systems
- speech recognition technology
- multi channel
- keyword spotting
- speaker recognition
- neural network
- cepstral coefficients
- speech recognition errors
- isolated word
- database systems