Principled Comparisons for End-to-End Speech Recognition: Attention vs Hybrid at the 1000-Hour Scale.
Aku RouheTamás GrószMikko KurimoPublished in: IEEE ACM Trans. Audio Speech Lang. Process. (2024)
Keyphrases
- end to end
- speech recognition
- hidden markov models
- language model
- speech synthesis
- speech processing
- speech signal
- pattern recognition
- automatic speech recognition
- speaker identification
- noisy environments
- speech recognizer
- speech recognition technology
- isolated word
- speech recognition systems
- congestion control
- speech recognizers
- neural network
- image coding
- maximum likelihood
- speech retrieval