In voice use technology, there is fundamental frequency as an important speech parameter. The use of voice use technology under noisy environment is also considered. Therefore, the technique to perform highly precise fundamental frequency estimation under noisy environment is searched for, and it has succeeded in many researches. In the fundamental frequency extraction under noise environment, an external noise is added to the input sound, this gives one of the factors which degrades the extraction accuracy. Then, the technique of performing highly precise fundamental frequency estimation under noisy environment by using the bone conduction sound was proposed, in which the noise is easily mixed in the input signal. In this research, in order to perform still more highly precise fundamental frequency estimation under noise environment, we propose a fundamental frequency estimation method in which properties of bone-conducted speech and air-conducted speech are combined.%音声利用における重要な音声パラメータとして基本周波数がある.音声利用には雑音環境下における利用も考えられる.そのため，雑音環境下においても高精度な基本周波数抽出を行える手法が求められており，現在も多くの研究がなされている.雑音環境下における基本周波数抽出では，入力音声に外部雑音が加わることが抽出精度を劣化させる要因の一つとなっている.そこで,雑音の混入しにくい骨導音声を入力音声として用いることで,雑音環境下においても高精度な基本周波数抽出を行う手法が過去に提案された.本研究では,雑音環境下においてさらに高精度な基本周波数抽出を行うために，骨導音声と気導音声の特性を組み合わせた基本周波数抽出法を提案する.
展开▼
