Should We Always Separate?: Switching Between Enhanced and Observed Signals for Overlapping Speech Recognition.
Hiroshi SatoTsubasa OchiaiMarc DelcroixKeisuke KinoshitaTakafumi MoriyaNaoyuki KamoPublished in: Interspeech (2021)
Keyphrases
- speech recognition
- noisy environments
- speech signal
- hidden markov models
- speech understanding
- language model
- automatic speech recognition
- speech processing
- signal processing
- speech synthesis
- speech recognizer
- speech recognition systems
- speaker identification
- pattern recognition
- speech recognition technology
- spectral analysis
- speech recognizers
- keyword spotting
- data mining
- speaker independent
- feature space