TalkingDroid

93 Followers
89 Following
677 Posts

I have this great idea for a new AI chatbot product, but with a twist: you don't have to prompt it.

And it gives, like, this really thorough, well-researched answer. And it's instant! As soon as you open it, the entire text is already generated, along with any figures or illustrations that are needed.

And to avoid hallucinations, all responses will have been compiled by human experts.

I call it a "book."

TGSpeechBox 3.0 is out. After seven betas, two release candidates, and more commits than I care to count — I'm calling it done.
I've been building this synthesizer for 5 months now, and 3.0 is the first release where I feel like it's genuinely a different piece of software from what came before. If you put 2.99 and 3.0 side by side, the difference isn't subtle. The vowels, the diphthongs, the stops, the prosody, the way connected speech actually flows, it's night and day. The Fujisaki pitch model not being a mechanical bull, the diphthong collapse system, the dictionary system fully done, the prominence and multiple-pitch pass pipeline. All of it came together in this cycle.
Every platform got real work. Linux is a first-class citizen now! A native Speech Dispatcher module, proper installer, PipeWire and ALSA auto-detection. No more pipes, no more shimmer. Android is on Google Play. iOS and macOS are on the App Store. Windows SAPI has a full settings UI. The pronunciation dictionary system ships on all of them.
None of this happened alone. This release belongs to the testers, the issue reporters, the dictionary contributors, and everyone who sent feedback across the betas. You shaped it.
3.0 is a milestone. I hope you hear it.
https://apps.apple.com/us/app/tgspeechbox/id6759512621
https://play.google.com/store/apps/details?id=com.tgspeechbox.tts
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300/TGSpeechBox-300.nvda-addon
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300/TGSBPhonemeEditor-v300.zip
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300/TGSpeechSapiSetup-v300.exe
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300/TGSpeechBox-v300.apk
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300/tgspeechbox-linux-x86_64-v-300.tar.gz
https://github.com/tgeczy/TGSpeechBox/releases/download/v-300/tgspeechbox-linux-aarch64-v-300.tar.gz
https://play.google.com/apps/testing/com.tgspeechbox.tts
https://testflight.apple.com/join/Y8RBtGBY
TGSpeechBox App - App Store

Download TGSpeechBox by Tamas Geczy on the App Store. See screenshots, ratings and reviews, user tips, and more apps like TGSpeechBox.

App Store
I got my agiga echo smart glasses today. Nothing like some seriously solid beta testing there in the app. You can tell that the majority of their testers were iOS. The android version is missing the signup button. Only got a login button.
Can the AI haters give it a rest already? Yes, I know there are concerns, but as a person with a disability, if I didn’t use every tool that was out there because I had concerns about it, I wouldn’t use anything. All this AI hatred is just cutting off our nose to spite our face.
Setting up my new phone. Back to Android after almost four years on iOS, now with a Pixel 10 Pro.
I gave Pocket TTS a shot. It's a local AI voice that claims to run well on the CPU, so it's faster to generate output and doesn't require a GPU. The main problem with it is the same as with so many voices these days, AI or otherwise: punctuation. Periods and commas sound the same, and question marks aren't always noticeable. Parenthetical phrases sound odd, and the double dash isn't handled at all. I love seeing this field continue to move forward, but I have yet to meet an AI voice I like.

Seeking recommendations for a reliable, accessible platform to host a support/discussion email list/listserv/snapshot of the 90's with around 1000 members. We're with GroupsIO right now but I'm tired of their in-browser inaccessibility. Cheap would be brilliant as my group doesn't directly make money, can pay something for solid service, accessibility and responsive support though.

I probably don't fancy self-hosting this much traffic.

Boosts appreciated.

Are you a photographer? Did you just take the most amazing picture of your life and are excited to share it on social media so others can bask in it's glory? If so, I feel that way about music, so I get it.
The question for you though, is did you add alt-text? Nope, this isn't going to be 'one of those posts' where I berate you for not doing so, it's actually to make you stop and think.
If the answer's yes, then this won't apply to you, but if the answer's no, let's see why not.

Is it because you think alt-text doesn't matter and nobody reads it anyway?
Is it because you don't know how to describe it so you just don't?

If it's either of those, let me see if I can help a bit.

Scenario 1, imagine how you'd explain your amazing picture to someone who can't see.
This might make you clam up because that seems like a really intimidating task, so let's not think of it like that. Instead, let's go for scenario number 2.

Imagine explaining it to someone on the phone that you just called up in huge excitement because of this crazy-good picture, they can't see it so you're just using natural language to explain why it has rocked your world.

do that, and then type what you think into the alt-text box.
Don't over-think it, you don't even have to go into gobs and gobs of detail if you'd rather not, but enough so that someone talking about your picture to someone else who hasn't seen it, will still be able to get all the high-points you yourself would want them to get, when talking about it.

If you do that for your pictures, people like me with no sight, and even your mum, dad, granny on the phone who isn't on social and won't join the family chat because they don't have a smart phone, will be able to enjoy it to, based on your personal description.

I'm not here to patronise you by any means, more to help you help others to enjoy your work even more, that's all.

#AltText helps so many people, not just those with a visual impairment.

If you made it this far, thank you.