FiLM Conditioning with Enhanced Feature to the Transformer-based End-to-End Noisy Speech Recognition.
Da-Hee YangJoon-Hyuk ChangPublished in: INTERSPEECH (2022)
Keyphrases
- end to end
- speech recognition
- noisy environments
- hidden markov models
- speech processing
- speech synthesis
- automatic speech recognition
- cepstral coefficients
- speech signal
- pattern recognition
- congestion control
- language model
- speech recognition technology
- speech recognizer
- speech recognition systems
- speaker identification
- speaker independent
- speech recognizers
- feature vectors
- computer vision
- isolated word
- signal to noise ratio
- feature set
- speaker dependent
- audio visual speech recognition