Segment-level Training of ANNs Based on Acoustic Confidence Measures for Hybrid HMM/ANN Speech Recognition.
S. Pavankumar DubaguntaMathew Magimai-DossPublished in: ICASSP (2019)
Keyphrases
- speech recognition
- artificial neural networks
- acoustic models
- hidden markov models
- speaker independent
- confidence measures
- speech recognition systems
- using artificial neural networks
- feed forward neural networks
- speech recognizers
- wall street journal corpus
- evolutionary artificial neural networks
- speech recognizer
- isolated word
- language model
- automatic speech recognition
- benchmark classification problems
- back propagation
- speech signal
- ground truth
- stereo vision
- noisy speech
- speech synthesis
- speech processing
- ann models
- confidence measure
- pattern recognition
- neural network
- noisy environments
- keyword spotting
- genetic algorithm
- speech recognition technology
- optical flow
- feedforward artificial neural networks
- handwriting recognition
- speaker dependent
- machine learning
- speaker identification
- training process
- computer vision