Detail výsledku
Automatic Lecture Indexunt Using Voice Recognition
This work deals with the problem of automatic audio indexing based on large vocabulary continuous speech recognition (LVCSR). The paper describes the development of such system and discusses the issues associated with the character of the input data. The problem is related to automatic processing of recordings containing lectures of the Signals and Systems course given in the Czech language. Compared to other speech recordings, these lectures are unique for their contents, which gives an outline to the accurate system development. In addition, speaker and language adaptation techniques are implemented and tested on anotated part of recordings.
adaptation, ASR, automatic, Czech, dictionary, Hidden Markov Model, HMM, HTK, indexing, lecture, LVCSR
@misc{BUT192615,
author="Ondřej {Glembek}",
title="Automatic Lecture Indexunt Using Voice Recognition",
year="2005",
pages="54",
address="Brno",
url="http://www.fit.vutbr.cz/~glembek/papers/dpi.pdf"
}