Wideband speech coding robust against packet loss
Takeshi Mori, Hitoshi Ohmuro, Yusuke Hiwasaki, Sachiko Kurihara, Akitoshi Kataoka · Electronics and Communications in Japan (Part III Fundamental Electronic Science) · 2006
This paper describes a wideband speech coding method that is robust against packet loss, and a packet transmission control method. The coding is based on 4-kHz-band CELP speech coding and consists mainly of multilevel coding using MDCT transform encoding, with a speech quality enhancement codec for the nonspeech portion and a higher subband enhancement codec for coding the higher subband at or above 4 kHz. This method can provide high-quality coding of nonspeech signals such as background music as well as wideband speech. If this coding method and packet transmission control are combined, speech transmission can continue without much deterioration even when the network becomes congested and packet loss occurs. Subjective and objective testing confirmed that, unlike conventional coding methods, high-quality speech transmission was possible even with 10% random packet loss when this method is used for speech transmission on an IP network. © 2006 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 89(12): 20–30, 2006; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.20291