An ASR Guided Speech Intelligibility Measure for TTS Model Selection.
Arun BabySaranya VinnaitherthanNagaraj AdigaPranav JawaleSumukh BadamSharath AdavanneSrikanth KonjetiPublished in: CoRR (2020)
Keyphrases
- model selection
- speech recognition
- text to speech
- automatic speech recognition
- cross validation
- speech synthesis
- speech signal
- selection criterion
- hyperparameters
- bayesian learning
- regression model
- noisy environments
- statistical learning
- sample size
- spontaneous speech
- meta learning
- parameter estimation
- mixture model
- statistical inference
- generalization error
- error estimation
- machine learning
- model selection criteria
- prosodic features
- information criterion
- parameter determination
- similarity measure
- marginal likelihood
- bayesian information criterion
- automatic model selection
- gaussian process
- motion segmentation
- hidden markov models
- feature selection
- variable selection
- generalization bounds
- signal to noise ratio
- hypothesis tests
- language model
- active learning