Otimização dos Codificadores VSELP e EFR por Refinamento da Modelagem Autoregressiva

Irene Heleonora S. P. Fantini, Luís Geraldo P. Meloni

DOI: 10.14209/sbrt.2000.5150140

Evento: XVIII Simpósio Brasileiro de Telecomunicações (SBrT2000)

Keywords: Codificação de Fala EFR VSELP ACELP TDMA PSQM

Abstract

"This paper extends the use of the refined autoregressive modeling to the EFR coder. The refined redesign is based on the observation that a better formant estimation is obtained by the use of a frame length multiple of pitch period and a synchronous frame position to glottal opening and closure. Simulations using telephone speech signals have shown that it is possible to obtain coding gains up to 3.5 dB with the segmental signal-to-noise ratio. The decoder does not require any modification and yet offers a better speech quality. The paper also compares the VSELP and EFR performances for both original and refined coders using segmental SNR and PSQM measures. "

Download