Thesis (Ed.M.)--Boston University

This thesis introduces a tool for analysis of audiovisual records. The tool uses the audio and closed captions supplied by the user to prepare text annotation. The annotation contains a transcript of the show which is based on the closed captions. In addition, speaker diarization is performed to mark who spoke when. The diarization is performed by a third party library. The library is evaluated on data from DIALOG corpus. The inner workings of the library are described. To assign the right portions of the text to the right section of the record Kaldi, a speech recognition toolkit, is used. Furthermore the thesis contains an overview describing how closed captions are created; overview of speech corpora creation; and a brief review of literature on record analysis. 1

Thesis (Ed.D.)--Boston University

Typically, instructional designers introduce audio in multimedia environments when a) it appears to be necessary -- for example, to provide feedback; or b) accessibility, availability, and/or hardware issues are insignificant -- in other words, It's there -- -- It sounds good -- It should work -- Why not use it? However, rarely is a decision to use audio based upon a thorough understanding of why, from a learning perspective, it is appropriate or optimal to do so. Clearly a lack of such an understanding can, and often does, lead to the inappropriate use of audio as a useful instructional medium. This study is designed: (1) to investigate the educational value of compressed speech in a multimedia environment, and (2) to evaluate how rate of speech may influence learning in a multimedia setting. It is also concerned with ascertaining whether: a) the recall of information at various levels of compressed speech decreases when audio delivery-rate (words-per-minute) increases; b) there is an interaction between task completion time and recall of information at various levels of compression; and c) relative to recall, there is an interaction between audio delivery-rate and increasing exposure to an audio stimulus. One hundred and ninety-two undergraduate students enrolled in business courses in two southeastern regional universities located in North Carolina and Virginia were randomly assigned to one of three groups. The three experimental test groups were normal speech, 125 words per minute, compressed speech: 175 words per minute and 200 words per minute. In all groups, participants were asked to listen to the solution of the puzzle, given in one of the three presentation rates of speech and they were given three different puzzles to solve. All data were collected at each assigned computer workstation. Data analysis revealed a difference between the three treatment groups (125 wpm, 175 wpm, 200 wpm) in recall scores due to the rate of compression, no interaction between completion times and the recall of information based on compression rate, and a difference in recall scores between the three treatment groups and the amount of exposure to the audio stimulus. / Ph. D.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.