Software & Resources

This page lists selected software, tools, and resources related to my research activities.

Some of these resources are actively maintained, while others are historically important software packages and research outputs from earlier work on statistical speech synthesis, speech signal processing, and related areas.

Speech Synthesis and Speech Processing

HTS: HMM/DNN-based Speech Synthesis System

HTS is a software toolkit for HMM/DNN-based speech synthesis. It has been widely used for research on statistical parametric speech synthesis.

SPTK: Speech Signal Processing Toolkit

SPTK is a suite of speech signal processing tools for UNIX environments, including tools for speech analysis, parameter conversion, synthesis filtering, vector quantization, and related signal processing tasks.

The newer SPTK 4.0+ series is a redesigned C++ implementation of SPTK. A differentiable PyTorch-based version, diffsptk, is also available for integrating classical speech signal processing modules into neural network models.

These recent versions have been developed in my research group. I have been involved in the research direction, while the software design and implementation have been led mainly by Takenori Yoshimura.

hts_engine API

hts_engine API is a speech synthesis engine API for generating speech waveforms from models trained by HTS.

Open JTalk

Open JTalk is a Japanese text-to-speech system based on HTS technology.

Sinsy

Sinsy is a singing voice synthesis system that generates singing voices from musical scores.

Multimodal Interaction

MMDAgent

MMDAgent is a toolkit for building voice interaction systems using CG characters. Source code and downloads are also available from SourceForge.

A successor system, MMDAgent-EX, is currently being developed and maintained separately by Prof. Akinobu Lee.

Historical Resources

Older software, demonstration pages, slides, and related materials are available in the old website archive.

Notes

The availability and maintenance status of these resources differ by project. Please refer to each project page for the latest information, documentation, and downloads.