Introduction

Microsoft wants us to throw away working computers for (opinionated) security benefits. Most of people only want to browse the Web, read their e-mails, but more importantly, leverage the last computer they bought while it’s still working : it’s good for the planet and their expenses. Software shouldn’t be the limitation here.
Their risk assessment (which actually doesn’t exist, because they have only few assets) couldn’t care less about whether or not their motherboard or CPU offers a TPM 2.0. They only want a secure operating system to be protected from trivial malware, a secure Web browser to execute megabytes of third-party JavaScript, and a secure e-mail client to handle all the spams they receive.

So here is yet another good time for most of humans to part away from EEE USA-giants, and opt for open (and free) alternatives (as GNU/Linux).

GNOME Orca default voice

Among all humans, some are part of my family (no kidding !), for which I usually play the “IT technician” role

My grand-father is highly visually impaired, as I previously wrote about when documenting how he used Android, and has been really keen to try what screen reading feature Linux could offer, following Windows replacement.
Windows… an OS he couldn’t even use for the last decade due to successive UX breaking changes (XP, 7, 10) he couldn’t adapt to, coupled to poor accessibility tools.

Orca screen reader is integrated to GNOME desktop environment by default, and uses speech-dispatcher (speechd) as an interface between screen readers and TTS (text-to-speech) engines for audio transcription.

On a default Debian 13 (Trixie) install, the screen reading feature software flow looks like this :

Screen text content --> Orca --> speechd --> eSpeak NG --> Audio output

Unfortunately, it turns out default (and open-source) TTS engine eSpeak NG output voices were absolutely horrendous, and we couldn’t get a clue of what they actually said. So we decided to give a try to SVOX Pico TTS engine, which is (proprietary but) publicly appreciated by the French community.

As we couldn’t find an up-to-date/working documentation, here it is !

Switching to Pico TTS engine

First you need to add contrib and non-free components to your apt sources list, for instance :

deb http://deb.debian.org/debian trixie main contrib non-free
deb http://deb.debian.org/debian-security trixie-security main contrib non-free
deb http://deb.debian.org/debian trixie-updates main contrib non-free

Then you have to install Pico utilities as well as speech-dispatcher Pico module :

apt-get update
apt-get install -y libttspico-utils speech-dispatcher-pico

Eventually, we edited /etc/speech-dispatcher/speechd.conf to enable Pico TTS by default :

# ...
DefaultModule pico
# ...
AddModule "pico" "sd_pico" "pico.conf"
# ...
# Note : you can replace below "fr" by "de", "en", "es" or even "it" for Italian !
LanguageDefaultModule "fr" "pico"

Now you only have to restart GNOME to make Orca properly use new speech-dispatcher configuration (restarting speech-dispatcherd.service didn’t seem to be sufficient).

Conclusion

SVOX Pico only proposes a single (female) voice, which is clearer and more importantly… understandable.

And don’t forget : an up-to-date and secure Windows, is a Windows which you’ve eventually got rid of !