2nd Joint Workshop on Multimodal Interaction and Related Machine Learning Algorithms 2005 - Edinburgh

Searching Speech: A Research Agenda

author:Douglas Oard, Institute for Advanced Computer Studies, University of Maryland
published: Feb. 25, 2007, recorded: June 2005, views: 36

Turn off the lights

You might be experiencing some problems with Your Video player.

Slides

Slides
0:00	Searching Speech: A Research Agenda
1:20	Some Grid Use at Maryland
3:56	Expanding the Search Space
5:34	Indexable Speech
8:05	A Web of Speech?
11:08	The Need for Scalable Solutions
13:27	Some Spoken Word Collections
14:42	Indexing Options
16:40	Supporting “Intellectual Access”
20:47	Some Technical Challenges
25:35	Start Time Error Cost
27:21	Shoah Foundation Collection
28:39	Interview Excerpt
30:43	MALACH Languages
31:14	Observational Studies
31:17	Relevance Criteria
31:19	Topicality
32:00	Test Collection Design
32:24	Test Collection Design
32:53	CLEF-2005 CL-SR Track
33:43	Additional Resources
34:30	English ASR
35:17	Warsaw (Poland)
35:30	Segment duration (s)
35:33	Keywords vs. Segment duration
35:36	Nodes descending from parents of leaves
35:53	Spoken dates in release ASR
35:59	Current classifier performance:
36:47	An Example English Topic
37:15	5-level Relevance Judgments
37:44	Comparing Index Terms
38:15	Searching Manual Transcripts
38:46	Category Expansion
40:09	Rethinking the Problem
40:56	Activation Matrix
41:11	Training Data: 196,000 Segments
41:28	Preprocessing Training Data
41:41	Characteristics of the Problem
42:18	Modeling Location
42:33	A Class Model for People
42:49	Search
43:04	TITLE
43:30	Some Open Issues
43:58	Non-English ASR Systems
45:09	Planning for the Future
45:53	The CLEF CL-SR Team
46:24	More Things to Think About
48:16	Final Thoughts
50:28	For More Information
50:42	TITLE

Related content

Visitors who watched this lecture also watched...