The connected word recognition problem Problem definition: Given a fluently spoken sequence of words, how can we determine the optimum match in terms.

Slides:



Advertisements
Similar presentations
DIJKSTRA’s Algorithm. Definition fwd search Find the shortest paths from a given SOURCE node to ALL other nodes, by developing the paths in order of increasing.
Advertisements

A Survey of Botnet Size Measurement PRESENTED: KAI-HSIANG YANG ( 楊凱翔 ) DATE: 2013/11/04 1/24.
Dynamic Programming Nithya Tarek. Dynamic Programming Dynamic programming solves problems by combining the solutions to sub problems. Paradigms: Divide.
Final presentation Final presentation Tandem Cyclic Alignment.
ECE 8443 – Pattern Recognition Objectives: Elements of a Discrete Model Evaluation Decoding Dynamic Programming Resources: D.H.S.: Chapter 3 (Part 3) F.J.:
電腦視覺 Computer and Robot Vision I Chapter2: Binary Machine Vision: Thresholding and Segmentation Instructor: Shih-Shinh Huang 1.
ETRW Modelling Pronunciation variation for ASR ESCA Tutorial & Research Workshop Modelling pronunciation variation for ASR INTRODUCING MULTIPLE PRONUNCIATIONS.
HMM-BASED PATTERN DETECTION. Outline  Markov Process  Hidden Markov Models Elements Basic Problems Evaluation Optimization Training Implementation 2-D.
Slide 1 EE3J2 Data Mining EE3J2 Data Mining Lecture 14: Introduction to Hidden Markov Models Martin Russell.
Approximate Dynamic Programming for High-Dimensional Asset Allocation Ohio State April 16, 2004 Warren Powell CASTLE Laboratory Princeton University
ITCS 6010 Spoken Language Systems: Architecture. Elements of a Spoken Language System Endpointing Feature extraction Recognition Natural language understanding.
Slide 1 EE3J2 Data Mining EE3J2 Data Mining - revision Martin Russell.
CMPT-825 (Natural Language Processing) Presentation on Zipf’s Law & Edit distance with extensions Presented by: Kaustav Mukherjee School of Computing Science,
String Matching String matching: definition of the problem (text,pattern) depends on what we have: text or patterns Exact matching: Approximate matching:
1 7-Speech Recognition (Cont’d) HMM Calculating Approaches Neural Components Three Basic HMM Problems Viterbi Algorithm State Duration Modeling Training.
So far: Historical introduction Mathematical background (e.g., pattern classification, acoustics) Feature extraction for speech recognition (and some neural.
1 TEMPLATE MATCHING  The Goal: Given a set of reference patterns known as TEMPLATES, find to which one an unknown pattern matches best. That is, each.
Abstract Developing sign language applications for deaf people is extremely important, since it is difficult to communicate with people that are unfamiliar.
7-Speech Recognition Speech Recognition Concepts
Dynamic Programming.
1 CS 552/652 Speech Recognition with Hidden Markov Models Winter 2011 Oregon Health & Science University Center for Spoken Language Understanding John-Paul.
Design and Implementation of Speech Recognition Systems Spring 2014 Class 13: Training with continuous speech 26 Mar
1 CPSC 320: Intermediate Algorithm Design and Analysis July 28, 2014.
8.0 Search Algorithms for Speech Recognition References: of Huang, or of Becchetti, or , of Jelinek 4. “ Progress.
Mobile Agent Migration Problem Yingyue Xu. Energy efficiency requirement of sensor networks Mobile agent computing paradigm Data fusion, distributed processing.
6/10/01Network Problems: DJK1 Network Problems Chapters 9 and 10.
ECE 8443 – Pattern Recognition ECE 8527 – Introduction to Machine Learning and Pattern Recognition Objectives: Evaluation Decoding Dynamic Programming.
Ch 5b: Discriminative Training (temporal model) Ilkka Aho.
Performance Comparison of Speaker and Emotion Recognition
ECE 8443 – Pattern Recognition ECE 8527 – Introduction to Machine Learning and Pattern Recognition Objectives: Elements of a Discrete Model Evaluation.
Short term forecast of travel times on the Danish highway network based on TRIM data Klaus Kaae Andersen Thomas Kaare Christensen Bo Friis Nielsen Informatics.
Chapter 7 - Sequence patterns1 Chapter 7 – Sequence patterns (first part) We want a signature for a protein sequence family. The signature should ideally.
DNA, RNA and protein are an alien language
What Dynamic Programming (DP) is a fundamental problem solving technique that has been widely used for solving a broad range of search and optimization.
Quiz Week 8 Topical. Topical Quiz (Section 2) What is the difference between Computer Vision and Computer Graphics What is the difference between Computer.
Dynamic Programming (DP), Shortest Paths (SP)
Definition of the Hidden Markov Model A Seminar Speech Recognition presentation A Seminar Speech Recognition presentation October 24 th 2002 Pieter Bas.
1 7-Speech Recognition Speech Recognition Concepts Speech Recognition Approaches Recognition Theories Bayse Rule Simple Language Model P(A|W) Network Types.
November 15, 2005M Jones1 Track Lists in Level 2 Outputs from SLAM provide lists of tracks faster than the current path to SVT and Level 2 via XTRP Sparsify.
Dynamic Programming - DP December 18, 2014 Operation Research -RG 753.
Content Based Coding of Face Images
IMAGE PROCESSING RECOGNITION AND CLASSIFICATION
Shortest Path from G to C Using Dijkstra’s Algorithm
Dijkstra’s shortest path Algorithm
핵심어 검출을 위한 단일 끝점 DTW 알고리즘 Yong-Sun Choi and Soo-Young Lee
An overview of decoding techniques for LVCSR
Distributed voting application for handheld devices
Lecture 5 Dynamic Programming
Network Layer – Routing 1
Statistical Models for Automatic Speech Recognition
LECTURE 15: HMMS – EVALUATION AND DECODING
8.0 Search Algorithms for Speech Recognition
RAY TRACING.
Sharat.S.Chikkerur S.Anand Mantravadi Rajeev.K.Srinivasan
Unit Test Pattern.
Lecture 5 Dynamic Programming
Design and Analysis of Algorithms (07 Credits / 4 hours per week)
Globally Optimal Generalized Maximum Multi Clique Problem (GMMCP) using Python code for Pedestrian Object Tracking By Beni Mulyana.
© 2012 Elsevier, Inc. All rights reserved.
Unit-4: Dynamic Programming
LECTURE 14: HMMS – EVALUATION AND DECODING
CONTEXT DEPENDENT CLASSIFICATION
30% grade = class presentations
Connected Word Recognition
Dynamic Programming Search
Lecture 5 Dynamic Programming
The connected word recognition problem Problem definition: Given a fluently spoken sequence of words, how can we determine the optimum match in terms.
Presenter : Jen-Wei Kuo
Design and Analysis of Algorithms (04 Credits / 4 hours per week)
Presentation transcript:

The connected word recognition problem Problem definition: Given a fluently spoken sequence of words, how can we determine the optimum match in terms of a concatenation of word reference patterns?

To solve the connected word recognition problem, we must resolve the following problems

connected word recognition

connected word recognition

connected word recognition

connected word recognition

connected word recognition The alternative algorithms: Two-level dynamic programming approach Level building approach One-stage approach and subsequent generalizations

Two-level dynamic programming algorithm

Two-level dynamic programming algorithm

Two-level dynamic programming algorithm

Two-level dynamic programming algorithm Computation cost of the two-level DP algorithm is: The required storage of the range reduced algorithm is: e.g., for M=300, N=40, V=10, R=5 C=1,320,000 grid points And S=6600 locations for D(b,e)

Level Building algorithm

Computation Of The Level Building Algorithm

Implementation Aspects of Level Building Beginning range reduction – MT Global range reduction – ε Test pattern ending range – δEND Reference pattern uncertainty regions – δR1 , δR2

Integration of a Grammar Network

1 I 5 ONE 9 BOOKS 13 OLD 2 WANT 6 A 10 COAT 3 NEED 7 AN 11 COATS 4 THREE 8 BOOK 12 NEW  

Current Words Predecessor State Uses Level Levels 2 I 1 3 WANT 4 NEED 5 THREE 6 A 7 AN 8 ONE NEW OLD 9 9* BOOK, COAT 10 7,8,9 BOOKS, COATS 11

One-pass or One-state algorithm Also known as frame-synchronous level building (FSLB) method For each test frame the accumulated distance is calculated as:

One-pass or One-state algorithm

One-pass or One-state algorithm

One-pass or One-state algorithm The problem is that no mechanism is provided for Controlling the resulting string path. For incorporating the level, we write:

Incorporating multiple candidate strings

Using HMM in level building

Segmental k-means Training Algorithm