Hidden Information State System A Statistical Spoken Dialogue System M. Gašić, F. Jurčíček, S. Keizer, F. Mairesse, B. Thomson, K. Yu and S. Young Cambridge.

Slides:



Advertisements
Similar presentations
1 Using the HTK speech recogniser to analyse prosody in a corpus of German spoken learners English Toshifumi Oba, Eric Atwell University of Leeds, School.
Advertisements

Pseudo-Relevance Feedback For Multimedia Retrieval By Rong Yan, Alexander G. and Rong Jin Mwangi S. Kariuki
Statisical Spoken Dialogue System Talk 2 – Belief tracking CLARA Workshop Presented by Blaise Thomson Cambridge University Engineering Department
Atomatic summarization of voic messages using lexical and prosodic features Koumpis and Renals Presented by Daniel Vassilev.
The CUED Speech Group Dr Mark Gales Machine Intelligence Laboratory
Dialogue Modelling Milica Gašić Dialogue Systems Group.
Dialogue Policy Optimisation
Statistical Dialogue Modelling Milica Gašić Dialogue Systems Group.
On-line dialogue policy optimisation Milica Gašić Dialogue Systems Group.
Speech Recognition Part 3 Back end processing. Speech recognition simplified block diagram Speech Capture Speech Capture Feature Extraction Feature Extraction.
Hidden Markov Models. Room Wandering I’m going to wander around my house and tell you objects I see. Your task is to infer what room I’m in at every point.
15.0 Utterance Verification and Keyword/Key Phrase Spotting References: 1. “Speech Recognition and Utterance Verification Based on a Generalized Confidence.
Seminar on Spoken Dialogue Systems
December 2006 Cairo University Faculty of Computers and Information HMM Based Speech Synthesis Presented by Ossama Abdel-Hamid Mohamed.
Gaussian Processes for Fast Policy Optimisation of POMDP-based Dialogue Managers M. Gašić, F. Jurčíček, S. Keizer, F. Mairesse, B. Thomson, K. Yu, S. Young.
Shallow Processing: Summary Shallow Processing Techniques for NLP Ling570 December 7, 2011.
1 Hidden Markov Model Instructor : Saeed Shiry  CHAPTER 13 ETHEM ALPAYDIN © The MIT Press, 2004.
Natural Language Understanding
Introduction to Automatic Speech Recognition
1 7-Speech Recognition (Cont’d) HMM Calculating Approaches Neural Components Three Basic HMM Problems Viterbi Algorithm State Duration Modeling Training.
1 Robust HMM classification schemes for speaker recognition using integral decode Marie Roch Florida International University.
Classifying Tags Using Open Content Resources Simon Overell, Borkur Sigurbjornsson & Roelof van Zwol WSDM ‘09.
Supervisor: Dr. Eddie Jones Electronic Engineering Department Final Year Project 2008/09 Development of a Speaker Recognition/Verification System for Security.
Speech and Language Processing
Develop a fast semantic decoder for dialogue systems Capability to parse 10 – 100 ASR hypotheses in real time Robust to speech recognition noise Semantic.
7-Speech Recognition Speech Recognition Concepts
Machine Learning in Spoken Language Processing Lecture 21 Spoken Language Processing Prof. Andrew Rosenberg.
Cognitive User Interfaces: An Engineering Approach Machine Intelligence Laboratory Information Engineering Division Cambridge University Engineering Department.
Evaluation of SDS Svetlana Stoyanchev 3/2/2015. Goal of dialogue evaluation Assess system performance Challenges of evaluation of SDS systems – SDS developer.
Yazd University, Electrical and Computer Engineering Department Course Title: Advanced Software Engineering By: Mohammad Ali Zare Chahooki 1 Introduction.
Reinforcement Learning for Spoken Dialogue Systems: Comparing Strengths & Weaknesses for Practical Deployment Tim Paek Microsoft Research Dialogue on Dialogues.
Develop a fast semantic decoder Robust to speech recognition noise Trainable on different domains: Tourist information (TownInfo) Air travel information.
IRCS/CCN Summer Workshop June 2003 Speech Recognition.
Modeling Speech using POMDPs In this work we apply a new model, POMPD, in place of the traditional HMM to acoustically model the speech signal. We use.
Learning Automata based Approach to Model Dialogue Strategy in Spoken Dialogue System: A Performance Evaluation G.Kumaravelan Pondicherry University, Karaikal.
LML Speech Recognition Speech Recognition Introduction I E.M. Bakker.
Jun-Won Suh Intelligent Electronic Systems Human and Systems Engineering Department of Electrical and Computer Engineering Speaker Verification System.
1 Boostrapping language models for dialogue systems Karl Weilhammer, Matthew N Stuttle, Steve Young Presenter: Hsuan-Sheng Chiu.
Still Talking to Machines (Cognitively Speaking) Machine Intelligence Laboratory Information Engineering Division Cambridge University Engineering Department.
Intelligent Robot Architecture (1-3)  Background of research  Research objectives  By recognizing and analyzing user’s utterances and actions, an intelligent.
ECE 8443 – Pattern Recognition ECE 8527 – Introduction to Machine Learning and Pattern Recognition Objectives: Reestimation Equations Continuous Distributions.
Cognitive Systems Foresight Language and Speech. Cognitive Systems Foresight Language and Speech How does the human system organise itself, as a neuro-biological.
CHAPTER 8 DISCRIMINATIVE CLASSIFIERS HIDDEN MARKOV MODELS.
Speech Recognition with CMU Sphinx Srikar Nadipally Hareesh Lingareddy.
Presented by: Fang-Hui Chu Discriminative Models for Speech Recognition M.J.F. Gales Cambridge University Engineering Department 2007.
Automatic Speech Recognition A summary of contributions from multiple disciplines Mark D. Skowronski Computational Neuro-Engineering Lab Electrical and.
HMM-Based Speech Synthesis Erica Cooper CS4706 Spring 2011.
金聲玉振 Taiwan Univ. & Academia Sinica 1 Spoken Dialogue in Information Retrieval Jia-lin Shen Oct. 22, 1998.
Develop a fast semantic decoder for dialogue systems Capability to parse 10 – 100 ASR hypothesis in real time Robust to speech recognition noise Trainable.
Chapter 7 Speech Recognition Framework  7.1 The main form and application of speech recognition  7.2 The main factors of speech recognition  7.3 The.
Integrating Multiple Knowledge Sources For Improved Speech Understanding Sherif Abdou, Michael Scordilis Department of Electrical and Computer Engineering,
Statistical Models for Automatic Speech Recognition Lukáš Burget.
A Hybrid Model of HMM and RBFN Model of Speech Recognition 길이만, 김수연, 김성호, 원윤정, 윤아림 한국과학기술원 응용수학전공.
1 7-Speech Recognition Speech Recognition Concepts Speech Recognition Approaches Recognition Theories Bayse Rule Simple Language Model P(A|W) Network Types.
By: Nicole Cappella. Why I chose Speech Recognition  Always interested me  Dr. Phil Show Manti Teo Girlfriend Hoax  Three separate voice analysts proved.
Evolvable dialogue systems
The CUED Speech Group Dr Mark Gales Machine Intelligence Laboratory
Sentiment analysis algorithms and applications: A survey
The CU Speech Group Machine Intelligence Laboratory
Adversarial Learning for Neural Dialogue Generation
Hidden Markov Models (HMM)
Conditional Random Fields for ASR
Statistical Models for Automatic Speech Recognition
Spoken Dialog System.
Integrating Learning of Dialog Strategies and Semantic Parsing
Statistical Models for Automatic Speech Recognition
Voice Activation for Wealth Management
CPSC 503 Computational Linguistics
Hierarchical, Perceptron-like Learning for OBIE
What is Artificial Intelligence?
Presentation transcript:

Hidden Information State System A Statistical Spoken Dialogue System M. Gašić, F. Jurčíček, S. Keizer, F. Mairesse, B. Thomson, K. Yu and S. Young Cambridge University Engineering Department | {mg436, fj228, sk561, farm2, brmt2, ky219, SEMANTIC DECODER AUTOMATIC SPEECH RECOGNISER NATURAL LANGUAGE GENERATOR TEXT TO SPEECH SYNTHESISER Machine sentence Machine dialogue action N-best list of user utterances SPOKEN DIALOGUE SYSTEM N-best list of user dialog. actions N-best list of user dialog. actions ãuNãuN ãu1ãu1 ãu2ãu2 gKgK g1g1 g2g2 Observation N-best list of user dialogue actions User goal partitions built according to the ontology rules Dialogue history Grounding states Hypotheses Every possible combination of observation, user goal and dialogue history Belief state Distribution over hypotheses h M =(ã u 2,p 1,g K ) h 1 =(ã u 1,p 2,g 3 ) h 2 =(ã u 3,p 2,g 2 ) p1p1 p2p2 h4h4 h1h1 h2h2 h3h3 h5h5 PARTIALLY OBSERVABLE MARKOV DECISION PROCESS-BASED DIALOGUE MANAGER Belief state is mapped to a point in the summary space Summary Space āmām The region the point falls in has a summary action associated to it by the policy Machine summary action amam Additional information from belief state is added to summary action Machine dialogue action DIALOGUE STATE MAINTAINING ACTION SELECTION DIALOGUE MANAGER POLICY OPTIMISATION[1] Reinforcement learning with a simulated user Monte Carlo Control algorithm for grid-based learning For tourist information domain: - 100,000 dialogues needed to train the optimal policy - The optimal policy divided the summary space in 1500 regions POLICY ATK – LARGE VOCABULARY CONTINUOUS SPEECH RECOGNITION Acoustic model – Speaker independent HMM, trained on 39 hours of data Language model - Statistical trigram, trained on 80M words Output – Confidence scored N-best list SEMANTIC TUPLE CLASSIFIERS[2] utterance(u) dialogue act type (slot 1 =value 1, … ) Probabilistic SVM classifiers predicting: P(slot i =value i |u), P(dia_act_type|u) Trained on 8000 sentences HMM-BASED SYNTHESISER[3] wordphonemeacoustics Flite Globally-tied density HMM Trained on 1 hour of CMU ARCTIC data STATISTICAL GENERATION Currently handcrafted Future work – data driven approach to optimise naturalness and style based on context Acknowledgements This research was funded by the UK EPSRC under grant agreement EP/F013930/1 and by the EU FP7 Programme under grant agreement (CLASSIC project: References [1] M Gašić, S Keizer, F Mairesse, J Schatzmann, B Thomson, K Yu, and SJ Young. Training and Evaluation of the HIS POMDP Dialogue System in Noise. In SigDial, Columbus, Ohio, [2] F Mairesse, M Gašić, F Jurčíček, S Keizer, B Thomson, K Yu, and SJ Young. Spoken Language Understanding from Unaligned Data using Discriminative Classification Models. Submitted to ICASSP09. [3] K Yu, T Toda, M Gašić, S Keizer, F Mairesse, B Thomson, and SJ Young. Probabilistic Modelling of F0 in Unvoiced Regions in HMM Based Speech Synthesis. Submitted to ICASSP09. Update history