Paper Info
Title
Investigations on exemplar-based features for speech recognition towards thousands of hours of unsupervised, noisy data
Abstract
The acoustic models in state-of-the-art speech recognition systems are based on phones in context that are represented by hidden Markov models. This modeling approach may be limited in that it is hard to incorporate long-span acoustic context. Exemplar-based approaches are an attractive alter-native, in particular if massive data and computational power are available. Yet, most of the data at Google are unsupervised and noisy. This paper investigates an exemplar-based approach under this yet not well understood data regime. A log-linear rescoring framework is used to combine the exemplar-based features on the word level with the first-pass model. This approach guarantees at least baseline performance and focuses on the refined modeling of words with sufficient data. Experimental results for the Voice Search and the YouTube tasks are presented.
Year
DOI
Venue
2012
10.1109/ICASSP.2012.6288904
Acoustics, Speech and Signal Processing
Keywords
Field
DocType
hidden Markov models,speech recognition,Voice Search,YouTube task,acoustic model,exemplar-based feature,hidden Markov model,log-linear rescoring framework,noisy data,speech recognition,unsupervised data,Exemplar-based speech recognition,conditional random fields,speech recognition
Noisy data,Computer science,Speaker recognition,Artificial intelligence,Speech recognition hidden markov models,Voice search,Training set,Conditional random field,Pattern recognition,Speech recognition,Hidden markov models speech recognition,Hidden Markov model,Machine learning
Conference
ISSN
ISBN
Citations 
1520-6149 E-ISBN : 978-1-4673-0044-5
978-1-4673-0044-5
5
PageRank 
References 
Authors
0.45
7
4
Name
Order
Citations
PageRank
Georg Heigold153937.69
Patrick Nguyen22724179.13
Mitchel Weintraub343989.30
Vincent Vanhoucke44735213.63