NTT DoCoMo Develops Speech Encoding Technology; "Real Voice" Call Quality

Apr 9, 2007
Lili Mine and Takeyoshi Yamada, Nikkei Electronics
Demonstration of the technology
Demonstration of the technology
[Click to enlarge image]

NTT DoCoMo, Inc has developed a speech coding technology that provides high-quality phone calls with significantly fewer arithmetic operations than that required in the existing technique in collaboration with DoCoMo Communications Laboratories USA, Inc and unveiled it at Wireless Technology Park 2007 held on April 4 and 5 at the Pacifico Yokohama convention center in Kanagawa Prefecture, Japan. DoCoMo's hTc Z PDA handset was used to demonstrate the technology. It enables the transmission of voice in a bandwidth ranging from 50 Hz to 16 kHz at a bitrate of 38-48 kbps. The main feature of the technology is its small calculation amount. Specifically, the number arithmetic operations required to compress or uncompress voice signals is "about half that of existing methods," says Kei Kikuiri of Research Laboratories, NTT DoCoMo. Thus, the technology is well-suited for devices with low processing power, such as mobile phones.

Based on one of the characteristics of human auditory perception, i.e. "distortion in loud sounds is relatively imperceptible," the latest technology employs a common technique to reduce data amount itself in which the coding resolution is limited to the minimum in the frequency range of louder voices while smaller voices are coded with higher resolution. The company explains that it has also used its proprietary technique to reduce operations.