Punctuation Prediction in Spontaneous Conversations: Can We Mitigate ASR Errors with Retrofitted Word Embeddings?
Lukasz AugustyniakPiotr SzymanskiMikolaj MorzyPiotr ZelaskoAdrian SzymczakJan MizgajskiYishay CarmielNajim DehakPublished in: INTERSPEECH (2020)
Keyphrases
- conversational speech
- automatic speech recognition
- speech retrieval
- recognition errors
- spontaneous speech
- prediction error
- prediction accuracy
- speech recognition
- word error rate
- broadcast news
- prediction algorithm
- word recognition
- english text
- speech signal
- prediction model
- co occurrence
- neural network
- low dimensional
- vector space
- hidden markov models