NOISE IN HMM-BASED SPEECH SYNTHESIS ADAPTATION:
ANALYSIS, EVALUATION METHODS AND EXPERIMENTS

Reima Karhila, Ulpu Remes and Mikko Kurimo
Department of Acoustics and Signal Processing, Aalto University School of Electrical Engineering, Finland

The paper submitted to Journal of Selected Topics on Signal Processing, special issue on statistic parametric speech synthesis investigated the effects of noise in HMM-based speech synthesis. The research was done by artificially corrupting clean speech with noise from NOISEX-92 database and using the noisy data to a clean average voice to target speakers.

This diagram will be revised before 9th April to illustrate comparison of EMLLR and CSMAPLR adaptation: