University of Cambridge > > CUED Speech Group Seminars > Efficient Decoding with Generative Score-Spaces Using the Expectation Semiring

Efficient Decoding with Generative Score-Spaces Using the Expectation Semiring

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Catherine Breslin.

Speech recognisers are usually based on hidden Markov models (HMMs). These are finite-state models, with observation vectors conditionally independent given the state sequence. This assumption yields efficient algorithms, but it limits the power of the model. An alternative type of model that allows a wide range of features and dependence structures is a log-linear model. For example, it can use longer-span, variable-length features. For decoding continuous speech, however, the optimal combination of segmentation of the utterance into words and word sequence must be found. Features must therefore be extracted for each possible segment of audio. For many types of features, this becomes slow. In this talk, I will discuss how long-span features can be derived from the likelihoods of word HMMs. Standard adaptation techniques can then be used. Derivatives of the log-likelihoods, which break the Markov assumption, are appended. I will show how to decode with this specific class of model in quadratic time.

Followed by a mini poster session for papers that will be presented the following week at SLT :

“Policy optimisation of POMDP -based dialogue systems without state space compression”, Milica Gasic, Matthew Henderson, Blaise Thomson, Pirros Tsiakoulis and Steve Young

“Discriminative Spoken Language Understanding Using Word Confusion Networks”, Matthew Henderson, Milica Gasic, Blaise Thomson, Pirros Tsiakoulis, Kai Yu, and Steve Young

“N-Best Error Simulation for Training Spoken Dialogue Systems”, Blaise Thomson, Milica Gasic, Matthew Henderson, Pirros Tsiakoulis, and Steve Young

Refreshments will be provided.

This talk is part of the CUED Speech Group Seminars series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.


© 2006-2023, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity