MLP-ASR: Sequence-length agnostic all-MLP architectures for speech recognition.
Jin SakumaTatsuya KomatsuRobin ScheiblerPublished in: CoRR (2022)
Keyphrases
- speech recognition
- multilayer perceptron
- automatic speech recognition
- neural network
- hidden markov models
- language model
- artificial neural networks
- noisy environments
- speech signal
- speech synthesis
- pattern recognition
- speech processing
- speech recognizer
- speech understanding
- speech recognition technology
- word error rate
- speech retrieval
- radial basis function
- speaker identification
- isolated word
- speech recognizers
- spontaneous speech
- conversational speech
- acoustic models
- cepstral coefficients
- speaker dependent
- speaker independent
- mobile devices
- signal processing