In this paper we present a new cascading conditional random field based phrase break model for text-to-speech systems, trained on the speaker specific acoustic data that the text-to-speech voices are trained on. The training phase does not require any manually labeled phrase break tags, as these are derived directly from the speaker specific recordings used for building the synthetic voices. We present objective evaluations on various corpora, and show that the proposed model compares well with state-of-the-art data-driven phrase break models, with the added benefit of being in a unified framework.
Reference:
Louw, J.A. & Moodley, A. 2016. Speaker specific phrase break modeling with conditional random fields for text-to-speech. In: 2016 Pattern Recognition Association of South Africa and Robotics and Mechatronics International Conference, 30 November - 2 December 2016, Stellenbosch, South Africa
Louw, J. A., & Moodley, A. (2016). Speaker specific phrase break modeling with conditional random fields for text-to-speech. http://hdl.handle.net/10204/10966
Louw, Johannes A, and Avashlin Moodley. "Speaker specific phrase break modeling with conditional random fields for text-to-speech." (2016): http://hdl.handle.net/10204/10966
Louw JA, Moodley A, Speaker specific phrase break modeling with conditional random fields for text-to-speech; 2016. http://hdl.handle.net/10204/10966 .
Presented in: 2016 Pattern Recognition Association of South Africa and Robotics and Mechatronics International Conference, 30 November - 2 December 2016, Stellenbosch, South Africa. Due to copyright restrictions, the attached PDF file only contains the abstract of the full-text item. For access to the full-text item, please consult the publisher's website. While waiting for the post-print or published PDF document from the publisher
ResearchSpace
