Low-Delay Singing Voice Alignment to Text.
Àlex Loscos, Pedro Cano, Jordi Bonada · 1999
In this paper we present some ideas and preliminary results on how to move phoneme recognition techniques from speech to the singing voice to solve the low-delay alignment problem. The work focus mainly on searching the most appropriate Hidden Markov Model (HMM) architecture and suitable input features for the singing voice, and reducing the delay of the phonetic aligner without reducing its accuracy.