Improving Character Error Rate is Not Equal to Having Clean Speech: Speech Enhancement for ASR Systems with Black-Box Acoustic Models.
Ryosuke SawataYosuke KashiwagiShusuke TakahashiPublished in: ICASSP (2022)
Keyphrases
- error rate
- black box
- speech recognition
- word error rate
- automatic speech recognition
- noisy speech
- speech signal
- speech enhancement
- black boxes
- noisy environments
- acoustic models
- test set
- speech recognition systems
- broadcast news
- hidden markov models
- noise reduction
- speaker identification
- vocal tract
- test cases
- language model
- speaker independent
- handwriting recognition
- linear prediction
- single channel
- discriminative training
- speech synthesis
- background noise
- speech recognizer
- neural network
- image processing