Speech synthesis statistical model training device, speech synthesis statistical model training method, and computer program product
AssigneeKabushiki Kaisha Toshiba
InventorsMasatsune Tamura, Masahiro Morita
Life of this patent
Estimating the term…
Abstract
A speech synthesis model training device includes one or more hardware processors configured to perform the following. Storing, in a speech corpus storing unit, speech data, and pitch mark information and context information of the speech data. From the speech data, analyzing acoustic feature parameters at each pitch mark timing in pitch mark information. From the acoustic feature parameters analyzed, training a statistical model which has a plurality of states and which includes an output distribution of acoustic feature parameters including pitch feature parameters and a duration distribution based on timing parameters.
Claims7
A speech synthesis model training device comprising:
- one or more hardware processors configured to perform:
- storing speech data, and pitch mark information and context information of the speech data;
- and
- from the speech data, analyzing acoustic feature parameters at each pitch mark timing in the pitch mark information;
- and
- from the acoustic feature parameters analyzed, training a statistical model which has a plurality of states and which includes an output distribution of acoustic feature parameters including pitch feature parameters and a duration distribution based on timing parameters.
A speech synthesis model training method implemented by a computer, the method comprising:
- storing, in a memory unit, speech data, and pitch mark information and context information of the speech data;
- analyzing, from the speech data, acoustic feature parameters at each pitch mark timing in pitch mark information;
- and
- training, from the analyzed acoustic feature parameters, a statistical model which has a plurality of states and which includes an output distribution of acoustic feature parameters including pitch feature parameters and a duration distribution based on timing parameters.
A computer program product comprising a non-transitory computer-readable medium including a speech synthesis model training program that causes a computer to execute:
- storing, in a memory unit, speech data, and pitch mark information and context information of the speech data;
- analyzing, from the speech data, acoustic feature parameters at each pitch mark timing in pitch mark information;
- and
- training, from the analyzed acoustic feature parameters, a statistical model which has a plurality of states and which includes an output distribution of acoustic feature parameters including pitch feature parameters and a duration distribution based on timing parameters.
Dependencies are inferred from the claim text.