# Non-English speech synthesis
If you're like me, you've been on the English-language Internet for so long that English is how you think of computer things.
Sadly, the English Internet has blind spots. The most obvious one of those is that the Eglish Internet doesn't care about other languages. So if you go and ask the search engine about "Linux speech synthesizers", hoping to get a robot voice to say "jazda", you'll realize that the results have nothing to offer you.
Instead, you stop for a moment, and decide to go to the smaller, a bit backwater, non-English Internet, and you ask "syntezator mowy linux", knowing that search results are likely to be infested with non-free solutions which you don't want to see, and you might not even find a free one.
The results are just as bad as in the English search, but somehow still worse. There's the unmaintained, one-person effort, a tutorial for Festival which asks you to download the voice bank from a long-dead URL, and a bunch of links referring to espeak which sounds awful even for a robot voice when asked to go outside of its English comfort zone. The guides to installing an Android software aren't even worth mentioning.
But there's something sparkling in the mud. A university website links to RHVoice, which is even packaged for Debian – although I don't know why it's in the non-free repo, the license on the github page is listed as GPL. If it sounds like an easy solution, it almost is. Almost.
## RHVoice
I don't run Debian, so I install the software in a container.
```
podman run -ai debian:bookworm
```
Once inside, I install vim, and I install the package and some voices:
```
echo 'deb http://deb.debian.org/debian bookworm non-free" >> /etc/apt/sources.list
apt update
apt install rhvoice rhvoice-polish
```
Then, I tell it to say something:
```
echo "jazda" | RHVoice-test -p natan -o - > test.wav
```
This will place the recording in a file. On the host side, I retrieve it using the container ID:
```
podman cp 1ea7197c68f6:test.wav Documents/jazda.wav
```
That's it! The recording sounds okay, and I can put it alongside the IPA pronunciation on the web site.