/ASR-music

Content of the research project on ASR in music. This repository contain the Kaldi's recipe and the explanation of how to use it.

Primary LanguageShellGNU General Public License v3.0GPL-3.0

Author Gerardo Roa Dabike

Supervisor Jon Barker

Automatic Speech Recognition in Music

This repository contain the Kaldi's recipe for the Dissertation project and the explanation of how to use it.

The directory asr-music is the kaldi recipe

Acoustic Model

The lexicon used in this project was CMUdict

Languaje Model

The LM was constructed using lyrics discographies from several artist.

Database Corpus

The database used was ACOMUS1 corpus. The construction of this corpus can be founded in ACOMUS