On the Effect of Purely Synthetic Training Data for Different Automatic Speech Recognition Architectures.
Nick RossenbachBenedikt HilmesRalf SchlüterPublished in: CoRR (2024)
Keyphrases
- automatic speech recognition
- training data
- speech recognition
- speech signal
- word error rate
- acoustic models
- spoken words
- conversational speech
- hidden markov models
- speech retrieval
- noisy environments
- speech corpus
- data sets
- broadcast news
- recognition errors
- spontaneous speech
- word recognition
- acoustic features
- training set
- decision trees
- learning algorithm
- speaker adaptation
- speech sounds
- neural network
- image acquisition
- document retrieval
- information retrieval
- machine learning