japanese_tts_ros

japanese_text_to_speech

This is a ROS node that outputs Japanese text as audio.

I'm using Open JTalk.

install

It depends on Open JTalk and aplay commands.

 sudo apt install open-jtalk open-jtalk-mecab-naist-jdic hts-voice-nitech-jp-atr503-m001 alsa-utils

Please install it inside the catkin workspace and catkin_make .

How to use

Please launch japanese_text_to_speech node. This node becomes the server and outputs audio text received from the client.

 rosrun japanese_text_to_speech japanese_text_to_speech

This node is implemented using actionlib. A simple client sample is provided with test_client node.

 rosrun japanese_text_to_speech test_client

Switching the voice model

By default, male audio is used, but you can also switch to other audio.

Place the .htsvoice file in an appropriate location and set ~hts_voice_file rosparam.

 rosrun japanese_text_to_speech japanese_text_to_speech _hts_voice_file:='/path/to/htsvoice'

Mei & Takumi

For example, you can use the voice model of Mei & Takumi developed by Nagoya Institute of Technology.

Use the following command to obtain and expand the voice model.

wget https://sourceforge.net/projects/mmdagent/files/MMDAgent_Example/MMDAgent_Example-1.8/MMDAgent_Example-1.8.zip
unzip MMDAgent_Example-1.8.zip

You can use the audio model by specifying the file under Voice/ of the extracted file. For example, if you want to use the voice model of May's happy voice, add the following argument:

rosrun japanese_text_to_speech japanese_text_to_speech _hts_voice_file:= ' PATH_TO_UNZIPPED_FILES/MMDAgent_Example-1.8/Voice/mei/mei_happy.htsvoice '

Reference: https://www.rcnp.osaka-u.ac.jp/~kohda/linux/espeak.html