Transcription-Free Fine-Tuning of Speech Separation Models for Noisy and Reverberant Multi-Speaker Automatic Speech Recognition.
William RavenscroftGeorge CloseStefan GoetzeThomas HainMohammad SoleymanpourAnurag ChowdhuryMark C. FuhsPublished in: CoRR (2024)
Keyphrases
- automatic speech recognition
- speech signal
- speech recognition
- noisy environments
- fine tuning
- acoustic models
- word error rate
- broadcast news
- conversational speech
- vocal tract
- speaker identification
- hidden markov models
- acoustic features
- speech sounds
- recognition errors
- speech corpus
- machine learning
- speaker verification
- speech recognizers
- non stationary
- spontaneous speech
- speech segments
- language model
- multi modal
- linear prediction
- speech synthesis
- error rate
- word recognition
- handwriting recognition
- automatic transcription
- spoken words
- phoneme recognition
- compound words
- speaker adaptation
- speech recognition systems
- mel frequency cepstral coefficients
- fine tuned
- sound source