VOICE ENCODING DEVICE AND VOICE ENCODING METHOD

摘要

PROBLEM TO BE SOLVED: To realize a voice encoding processing which has less dependence on the precision of an noise model, which is resistant to a noise signal component and which has high quality even in a background noise environment. SOLUTION: A noise section detection part 101 separates an input signal into a voice section and a section except for the section and detects the signal in the section except for the voice section as background noise. A noise level estimation part 102 estimates the noise level of the noise section. An information source model storage part 103 stores an information source model obtained by modeling a parameter string with respect to a voice input signal which does not include noise. A voice analysis part 104 analyzes the input signal and extracts a parameter. A parameter quantizing part 105 quantizes the parameter extracted in the voice analysis part 104 based on the information source model and the noise level and outputs a code corresponding to a quantized value.