[PYTHON] Make Raspberry Pi speak Japanese using OpenJtalk

What you can do with this article

-Installed on OpenJtalk's Raspberry Pi • The actual sound output "Hello"

Operating environment

・ Raspberry Pi3 model B ・ OS: Raspbian

Premise

・ Can be operated with Raspberry Pi ssh connection or can be operated with display ・ Japanese input is possible

Install OpenJtalk

I installed it by referring to this article.

First, do the following

sudo apt-get update #Update
sudo apt-get install open-jtalk #Installation

Then also install the recommended packages.

sudo apt-get install open-jtalk-mecab-naist-jdic hts-voice-nitech-jp-atr503-m00

test

Now you can output audio! OpenJtalk is in the order of creating a Japanese audio file and then playing it. First, create an audio file below.

echo "Hello" | open_jtalk -x /var/lib/mecab/dic/open-jtalk/naist-jdic -m /usr/share/hts-voice/nitech-jp-atr503-m001/nitech_jp_atr503_m001.htsvoice -ow ~/ojtalk.wav

You should now have an audio file called ojtalk.wav in the same directory. Let's play next.

aplay ~/ojtalk.wav

"Hello" Was output?

Opinion of option

-x dictionary to use -m Which voice to use (male or female, etc.) -ow specify output file

When there is a lot of noise

By the way, if it is a 3.5mm analog output, noise may be added and it may not sound good. (I was ...) In that case, try connecting to the display with HDMI and outputting from the display. The noise, probably due to the circuit, disappears.

Output without saving the audio file

This time, I created an audio file and then output it, but if you feel a little uncomfortable that the audio file is saved every time you speak, you can do it without creating an audio file below.

echo "Hello" | open_jtalk -x /var/lib/mecab/dic/open-jtalk/naist-jdic -m /usr/share/hts-voice/nitech-jp-atr503-m001/nitech_jp_atr503_m001.htsvoice -ow /dev/stdout | aplay --quiet

Download female voice

Use the MMDAgent package.

First, execute the following to download

wget https://sourceforge.net/projects/mmdagent/files/MMDAgent_Example/MMDAgent_Example-1.7/MMDAgent_Example-1.7.zip

Then unzip the zip

unzip ./MMDAgent_Example-1.7.zip

And mei-chan's voice folder in the folder Copy it under the directory / usr / share / hts-voice /.

sudo cp -r ./MMDAgent_Example-1.7/Voice/mei/ /usr/share/hts-voice/

If you change the acoustic model to this, it will be a female voice. Optional -m acoustic model It is the part corresponding to.

echo "Hello"|open_jtalk -x /var/lib/mecab/dic/open-jtalk/naist-jdic -m /usr/share/hts-voice/mei/mei_normal.htsvoice -ow ~/ojtalk.wav
aplay ~/ojtalk.wav

I think this is a female voice.

Summary

This time, we have delivered the installation of OpenJtalk and its test. Next, I would like to create a program that talks about the information obtained by API and what was web scraped! Then!