An 8.3mW 1.6Msamples/s multi-modal event-driven speech enhancement processor for robust speech recognition in smart glasses.
Jinmook LeeSeongwook ParkInjoon HongHoi-Jun YooPublished in: ESSCIRC (2016)
Keyphrases
- multi modal
- speech recognition
- noisy environments
- event driven
- speech enhancement
- speech signal
- noisy speech
- hidden markov models
- speech synthesis
- language model
- speaker identification
- background noise
- automatic speech recognition
- vocal tract
- audio visual
- spectral subtraction
- pattern recognition
- additive noise
- spectral analysis
- neural network
- maximum likelihood
- principal component analysis
- machine learning