Three-Stage Error Concealment for Distributed Speech Recognition (DSR) with Histogram-Based Quantization (HQ) Under Noisy Environment.
Chia-Yu WanYi ChenLin-Shan LeePublished in: ICASSP (4) (2007)
Keyphrases
- noisy environments
- speech recognition
- error concealment
- speech signal
- video transmission
- automatic speech recognition
- coded images
- visual quality
- speaker verification
- packet loss
- language model
- speaker identification
- hidden markov models
- speech synthesis
- bitstream
- motion vectors
- audio visual speech recognition
- error resilient
- pattern recognition
- error resilience
- voice activity detection
- video quality
- background noise
- speech enhancement
- error propagation
- noisy speech
- wireless channels
- macroblock
- noise reduction
- bit rate
- computer vision
- low level
- neural network