Convolutional Variational Autoencoders for Spectrogram Compression in Automatic Speech Recognition.
Olga YakovenkoIvan BondarenkoPublished in: AIST (Supplement) (2020)
Keyphrases
- automatic speech recognition
- speech signal
- speech recognition
- word error rate
- restricted boltzmann machine
- hidden markov models
- denoising
- deep belief networks
- broadcast news
- conversational speech
- image compression
- spoken words
- image segmentation
- speech retrieval
- word recognition
- deep learning
- speech corpus
- recognition errors
- neural network
- noisy environments
- multiscale
- pattern recognition
- language model
- compression ratio
- unsupervised learning
- multiresolution
- probabilistic model
- machine learning
- speech sounds
- sparse coding