Repackagingaugment: Overcoming Prediction Error Amplification in Weight-Averaged Speech Recognition Models Subject to Self-Training.
Jae-Hong LeeDong-Hyun KimJoon-Hyuk ChangPublished in: ICASSP (2023)
Keyphrases
- speech recognition
- prediction error
- hidden markov models
- acoustic models
- speech processing
- language model
- speaker identification
- pattern recognition
- speech recognition technology
- noisy environments
- speech synthesis
- motion compensated
- automatic speech recognition
- audio visual speech recognition
- speech recognizers
- speech recognizer
- autoregressive
- motion vectors
- bit rate
- multiresolution