Voice Conversion Using Input-to-Output Highway Networks

• DOI: 10.1587/transinf.2017edl8034
• Publication Date: 2017-07-31T22:19:50Z
• Authors: Yuki SAITO; Shinnosuke TAKAMICHI; Hiroshi SARUWATARI
• Keywords: Smoothing; Cepstrum; Mel-frequency cepstrum

This paper proposes Deep Neural Network (DNN)-based Voice Conversion (VC) using input-to-output highway networks. VC is a speech synthesis technique that converts input features into output parameters, and DNN-based acoustic models for are used to estimate the parameters from parameters. Given often in same domain (e.g., cepstrum) VC, this networks connected output. The predict weighted spectral differentials between architecture not only alleviates over-smoothing effects degrade quality, but also effectively represents characteristics of experimental results demonstrate proposed outperforms Feed-Forward neural terms quality speaker individuality converted speech.