Name
Abstract | ||
|---|---|---|
This paper introduces a novel combination of Artificial Neural Networks (ANNs) and Hidden Markov Models (HMMs) for Automatic SpeechRecognition (ASR), relying on ANN non-parametric estimation of the emission probabilities of an underlying HMM. A gradientascent global training technique aimed at maximizing the likelihood (ML) of acoustic observations given the model is presented. A maximum aposteriori variant of the algorithm is also proposed as a viable solution to the "divergence problem" that may arise in the ML setup. A 46.34% relative word error rate reduction withresp ect to standard HMMs was obtained in a speaker-independent, continuous ASR task witha small vocabulary. |
Year | DOI | Venue |
|---|---|---|
2001 | 10.1007/3-540-44668-0_81 | ICANN |
Keywords | Field | DocType |
hidden markov models,acoustic observation,continuous asr task witha,automatic speechrecognition,standard hmms,artificial neural networks,markovian hybrid model,ml setup,emission probability,ann non-parametric estimation,divergence problem,robust connectionist,continuous speech recognition,automatic speech recognition,word error rate,artificial neural network,hidden markov model | Markov process,Divergence problem,Computer science,Artificial intelligence,Artificial neural network,Connectionism,Pattern recognition,Markov model,Word error rate,Speech recognition,Hidden Markov model,Vocabulary,Machine learning | Conference |
Volume | ISSN | ISBN |
2130 | 0302-9743 | 3-540-42486-5 |
Citations | PageRank | References |
4 | 0.55 | 4 |
Authors | ||
2 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Edmondo Trentin | 1 | 286 | 29.25 |
| Marco Gori | 2 | 4 | 0.55 |