SuperLectures.com

DISCRIMINATIVE DURATION MODELING FOR SPEECH RECOGNITION WITH SEGMENTAL CONDITIONAL RANDOM FIELDS

Full Paper at IEEE Xplore

Speech Analysis

Presented by: Patrick Nguyen, Author(s): Justine Kao, Stanford University, United States; Geoffrey Zweig, Patrick Nguyen, Microsoft Research, United States

This paper describes a new approach to modeling duration for LVCSR using SCARF, a toolkit for speech recognition with segmental conditional random fields. We utilize SCARF’s ability to integrate long-span, segment-level features to design and test duration models that help discriminate between correct and incorrect word hypotheses. We show that the duration distributions of correct and incorrect word hypotheses differ. Given a word hypothesis in the lattice and its duration, conditional length probabilities are integrated to the SCARF system as duration features. We evaluate three kinds of duration features on Broadcast News: word, pre- and post-pausal durations, and word span confusions. Adding the duration features to SCARF results in an up to 0.3% improvement over a state-of-the-art discriminatively trained baseline of 15.3% WER on a Broadcast News task.


  Speech Transcript

|

  Slides

Enlarge the slide | Show all slides in a pop-up window

0:00:16

  1. slide

0:00:30

  2. slide

0:01:32

  3. slide

0:03:34

  4. slide

0:08:14

  5. slide

0:09:21

  6. slide

0:10:00

  7. slide

0:11:15

  8. slide

0:12:09

  9. slide

0:13:01

 10. slide

  Comments

Please sign in to post your comment!

  Lecture Information

Recorded: 2011-05-25 10:30 - 10:50, Panorama
Added: 15. 6. 2011 16:48
Number of views: 30
Video resolution: 1024x576 px, 512x288 px
Video length: 0:15:07
Audio track: MP3 [5.08 MB], 0:15:07