Recent searches
Search options
In FLOSS projects we tend to accept to take responsibility for what we do, but we tend to reject responsibility for what we don't do, as if not doing it wasn't a choice, conscient or not. Choosing bit to work on something is still doing something, and we really need to learn to take responsibility for that. Collectively and individually. I am as responsible of the sore state of text-to-speech on FLOSS ecosystems like GNOME as everyone else.
What I'm trying to say is: we can't accept that accessibility is just for the disabled to develop. Commons have to be developed and maintained collectively by everybody, and we should particularly develop things we don't personally need to use the commons so others can be part of them too.
Yeah I once again tried text-to-speech on GNOME and my hears got quickly shredded by espeak.
I don't know how but I missed Eitan Isaacson's presentation of Spiel at GUADEC 2024. https://youtu.be/xseIsaxrlXo Eitan explains very well what the problems are with voice synthesis on Linux, how his work on Spiel can help solve part of the issue them, and that the missing bits are: distributing voice models. We're not there yet, but I'm glad to see such progress! I couldn't just rant without showing what I missed. 
@KekunPlazas re text to speech, have you tried spiel? Iiuc orca can optionally use it as well?
@philn No, IIUC from what Joannie says here, Orca relies on speech-dispatcher, I assume it has a Spiel plugin? I found plugins for Festival and Flite on Fedora. I tried the Flite one, which gave me barely better result than espeak. I'm trying to understand the state of TTS in general, but finding information isn't easy. https://gitlab.gnome.org/GNOME/orca/-/issues/378
@KekunPlazas @philn my personal recommendation is Pied, which is a simple Flatpak to install Piper voice models, it's working really well with the brazilian portuguese voices and the english ones are basically the best: https://github.com/Elleo/pied
@joojmachine @philn I can't find it on Flathub, do you have a reference to it? 
@KekunPlazas @philn It's not on Flathub (for now only, and I believe they could use some help, I believe the discussion stagnated when it comes to sandbox holes): https://github.com/Elleo/pied/issues/2
@joojmachine @philn Oh wow it's sooo much better now! It's night and day! Why the heck don't we have Piper in Speech Dispatcher by default‽
@KekunPlazas Sur les voix disponibles, j'ai posé la question aux gens d'Aciah.
je pense que nos attentes sur les voix ont été déformées par "Ok google", "siri", "alexa play despacito" et autres assistants... parce que leur réponse n'a pas été sur la qualité de la voix de synthèse.
Ce que je me pose vraiment la question est, comment avoir des nouvelles sur
Neutron.
https://blogs.gnome.org/a11y/2024/06/18/update-on-newton-the-wayland-native-accessibility-project/
On pourrait wild tag la personne, mais je pense que ça se fait pas.
Sorry english speakers, i do speak english, at least this is a good example of not being able of taking part of a discussion because the content is not accessible.
@rumi Si je dis pas de conneries Newton est un remplaçant d'AT-SPI, soit l'API qui permet au lecteur d'écran et aux applications de communiquer, et la synthèse vocale c'est ce que peut faire le lecteur d'écran avec les données recueillies depuis les applications, donc Newton ne changera rien à ça.
Si je dis pas de bêtises Newton est terminé et ne devrait pas avoir besoin d'évoluer ? Après je sais pas j'ai pas bien suivi.
@KekunPlazas c'est terminé et déjà sur mon ordinateur ? I use GNOME btw
@rumi Je sais pas, mais bon c'est une API donc y'a rien à installer normalement, c'est juste une façon de communiquer entre le lecteur d'écran, le compositeur et les applications, il suffit que ces composants implémentent et utilisent l'API. Enfin encore une fois si je dis pas de bêtises. En vrai j'en sais rien.