SYNT++: Utilizing Imperfect Synthetic Data to Improve Speech Recognition.
Ting-Yao HuMohammadreza ArmandpourAshish ShrivastavaJen-Hao Rick ChangHema KoppulaOncel TuzelPublished in: ICASSP (2022)
Keyphrases
- synthetic data
- speech recognition
- hidden markov models
- speech synthesis
- pattern recognition
- automatic speech recognition
- real image data
- speech recognizer
- speech processing
- real world
- speech recognition systems
- speech signal
- language model
- data sets
- keyword spotting
- noisy environments
- signal processing
- image processing
- speaker dependent
- speech recognizers
- feature vectors
- natural language
- speech understanding
- speaker adaptation