@techreport{oai:ipsj.ixsq.nii.ac.jp:00056563, author = {宮本, 大輔 and 中村, 圭吾 and 戸田, 智基 and 猿渡, 洋 and 鹿野, 清宏 and Daisuke, Miyamoto and Keigo, Nakamura and Tomoki, Toda and Hiroshi, Saruwatari and Kiyohiro, Shikano}, issue = {10(2009-SLP-075)}, month = {Jan}, note = {肉伝導音声変換はNon-Audible Murmur (NAM)マイクロフォンで収録される肉伝導音声の音質向上に効果的である.この手法では,肉伝導音声から空気伝導音声へ変換するための確率モデルが事前に学習される.肉伝導音声の音響特性は,NAMマイクロフォンの圧着位置などの収録環境に敏感であるため,実際の使用においては学習時と変換時の音響特性の不一致により,しばしば変換音質が大きく劣化する.この問題に対して,我々は肉伝導音声変換のためのCepstrum Mean Subtraction (CMS)とConstrained Structural Maximum A Posteriori Linear Regression (CSMAPLR),またはSignal Bias Removal (SBR)とCSMAPLRの組み合わせに基づく教師無しの音響特性補正法を提案する.実験結果から,提案手法により音響特性の不一致に起因する変換音質の劣化が大幅に低減されることを示す., Body transmitted voice conversion is very effective for enhancing body transmitted speech recorded with Non-Audible Murmur (NAM) microphone. In this method, a probabilistic model to convert body transmitted speech into natural speech is trained previously. Because acoustic characteristics of body transmitted speech is sensitive to recording conditions such as a location of NAM microphone, significant degradation of the conversion performance is often caused in practical situations by acoustic mismatches between the training and the conversion processes. To alleviate this problem, we propose unsupervised acoustic compensation methods based on combination of Cepstrum Mean Subtraction (CMS) and Constrained Structural Maximum A Posteriori Linear Regression (CSMAPLR), or combination of Signal Bias Removal (SBR) and CSMAPLR for body transmitted voice conversion. Experimental results demonstrate that the proposed methods significantly reduce the quality degradation of the converted speech caused by the acoustic mismatches.}, title = {音響特性補正の導入による肉伝導音声変換の収録環境適応}, year = {2009} }