Suppression of unmasked quantization noise for channel scalable 3D audio codec

2016 Fiscal Year Final Research Report

• Project/Area Number: 26420348
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Communication/Network engineering
• Research Institution: University of Toyama
• Principal Investigator: Ando Akio 富山大学, 大学院理工学研究部(工学), 教授 (00545668)
• Project Period (FY): 2014-04-01 – 2017-03-31
• Keywords: マルチチャネルオーディオ / 圧縮符号化 / スケーラブル伝送 / 空間音響 / 音響信号処理

Outline of Final Research Achievements

In a 8K super hi-vison test broadcasting, which started from 2016 in Japan, a 22.2 channel audio was adopted. However, such a multichannel audio is difficult to reproduce in a home and a new system that reproduces the original sound impression with a lesser number of channels is necessary. We developed a 8.1 channel system as such a method.
A scalable transmission in which a signal of the home reproduction system is embedded in a 22.2 channel signal usually requires a matrix conversion. However, if the converted signal is coded by an audio codec, the quantization noise that should be masked by the other sounds becomes audible after decoded and inversely converted. In this research, a noise compression method was developed and the audible noise was reduced by more than 20dB within a time delay of 340 ms.

Free Research Field

感性情報学、通信・ネットワーク工学　（音響工学）