Please use this identifier to cite or link to this item:
|Title:||Automatic speech transcription from DVD||Authors:||Ranjit Monisha Deva Belley||Keywords:||DRNTU::Engineering::Electrical and electronic engineering||Issue Date:||2013||Abstract:||Speech Transcription is a process of converting the speech into text. That is mapping a spoken language onto written symbols. Spoken language is a continuous phenomenon, made up of a potentially unlimited number of components. This is an application of Speech Recognition. Speech Recognition (SR) is a process to translate the speech into text format. The above process, when done automatically it is called as Automatic Speech Recognition (ASR). ASR is the challenging problems of modem man-kind. The speech recognizer should be trained with the transcribed data. Doing this process manually is expensive. Hence it is tried to be done automatically. In this project, the subtitles and its respective time information are extracted from DVD. Then the speech from its particular time information for its respective subtitle is taken from the audio information. There may be mismatch for the speech and the subtitle with the time information. Here the time domain methods are used to overcome this problem. That is done by taking energy for each time information. The proposed project does the extraction of the subtitles and the speech information for each time information from DVD automatically. The extraction of the subtitles can be performed by converting the graphical information into text information. The corresponding speech signal must also be segmented using the timing information. The information will then be converted into a format that is suitable for training a speech recognizer. This process is done automatically in MATLAB.||URI:||http://hdl.handle.net/10356/55245||Fulltext Permission:||restricted||Fulltext Availability:||With Fulltext|
|Appears in Collections:||EEE Theses|
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.