The voice layer for agents. Real-time, accurate, natural voice APIs: TTS, STT, turn-taking & voice design. Devs: build → https://nitter.gq/t.co/BMTX2dOALw
Joined September 2025
- Tweets317
- Following38
- Followers5.9K
- Likes610
Pinned Tweet
A new frontier for on-device speech.
We're open-sourcing Phonon, a suite of TTS models under 50m parameters, setting a new standard for naturalness and accuracy across languages, with the broadest inference support to date.
Phonon-31m is available under Apache 2.0.
gradium.ai/phonon🧵
Gradium retweeted
phonon by @GradiumAI is just such a good local text-to-speech model. imo noticeably better than kokoro.
with int8 weights it's only 44 MB!
on a fresh page load with the model cached you can have speech in one of many voices in literally less than the blink of an eye!
26x real time on an iphone. in safari. from a model that's under 40 MB brotli'd, with the same voices and quality as @GradiumAI's own runtime
every word is timestamped, so captions and read-alongs just work.
blixt.xyz/phonon/ (open source, apache-2.0)
Welcome to our weekly community spotlight.
Congratulations to the winners of the @active_builders x Gradium hackathon:
1st: Mirror Voice
2nd: Thread
3rd: Parley
Together, the three projects use Gradium voice cloning, speech-to-text, and TTS.
activebuilders.dev/hackathon…
2nd place: Thread by Andre Ani.
You think out loud, and Cara, Thread’s voice assistant, turns your ideas into a visual mind map. Gradium handles the speech-to-text and Cara’s spoken replies, so you can keep talking while the board grows.
Demo: share.descript.com/view/oFoz…
3rd place: Parley by @radiumcoders.
You hold a key and speak, and Parley sends a cleaned-up version of your prompt to Claude Code. Gradium transcribes your speech and reads Claude’s replies aloud. Claude Haiku cleans up the transcript before Parley sends it.
Demo: youtube.com/watch?v=AXHbLjNj…
Gradium is now HDS certified.
HDS is France’s certification for hosting health data. Healthcare teams can now build with Gradium: patient calls, appointment booking, medical transcription, and more.
Reach out to enable it on your account.
trust.gradium.ai
Our Maëlys just ranked #1 for French naturalness in Podonos’ new voice benchmark, Onepin.
Merci, Podonos.
Have you heard it yet?
gradium.link/maelys
You picked the best text-to-speech model, but it sounds bad. Do you know why?
Every TTS benchmark so far ranks models. But no one ships a model to their users. They ship a voice.
At Onepin, we've 𝐛𝐞𝐧𝐜𝐡𝐦𝐚𝐫𝐤𝐞𝐝 55,556 𝐯𝐨𝐢𝐜𝐞𝐬 across 11 languages, each one scored on its own. One pattern kept showing up in the data:
🔹 Two voices from the same model can be 0.67 points apart on naturalness, while the top two models sit just 0.04 apart.
🔹 In 10 of 11 languages, the #1 voice comes from a model the leaderboards don't rank first.
🔹 The best English (US) voice is @cartesia's Evan.
🔹 The best French voice is @GradiumAI's Maëlys.
🔹 The best German voice comes from @ElevenLabs.
🔹 A model-level average hides both: the spread between voices of one model, and the shift in winners from one language to the next.
So 𝐖𝐡𝐢𝐜𝐡 𝐓𝐓𝐒 𝐦𝐨𝐝𝐞𝐥 𝐢𝐬 𝐛𝐞𝐬𝐭? is the wrong question. The correct one is: 𝐖𝐡𝐢𝐜𝐡 𝐯𝐨𝐢𝐜𝐞 𝐰𝐨𝐫𝐤𝐬 𝐛𝐞𝐬𝐭 𝐟𝐨𝐫 𝐦𝐲 𝐥𝐚𝐧𝐠𝐮𝐚𝐠𝐞 𝐚𝐧𝐝 𝐦𝐲 𝐮𝐬𝐞 𝐜𝐚𝐬𝐞?
Today we're launching the Onepin Voice Benchmark, 𝐭𝐡𝐞 𝐰𝐨𝐫𝐥𝐝'𝐬 𝐟𝐢𝐫𝐬𝐭 𝐓𝐓𝐒 𝐯𝐨𝐢𝐜𝐞 𝐛𝐞𝐧𝐜𝐡𝐦𝐚𝐫𝐤 𝐩𝐞𝐫 𝐥𝐚𝐧𝐠𝐮𝐚𝐠𝐞.
If you're choosing a voice for a product, an AI agent, or content in more than one language, this is built for you.
Talking voice AI in person this month.
Nov 5, Paris: drinks with @covaldev and voice agent builders
luma.com/wszf3b7v
Nov 10, Lisbon: Alexandre Défossez with @AgoraIO on voice AI beyond the demo
luma.com/wqg33o39
Nov 11, Lisbon: Alexandre on stage at @WebSummit
Come say hi.
Gradium retweeted
This is not talked about enough but with @GradiumAI we released a chrome extension last week ! It can read, transcribe and translate any content you want. Pretty useful plus i had a lot of fun designing it!
Gradium will return to the Big Berlin Hack on November 28 and 29.
Meet our team and get free Gradium credits to add voice to your hackathon projects.
Register: gradium.link/big-berlin-hack
📣📣📣 Big Berlin Hack is back 🚀🚀🚀
And this time, even bigger:
🗓️ 28 & 29 November
👥 350+ builders from all over Europe
💰 Bigger prize pool
Last time we had >1500 applications for Big Berlin Hack, so better be fast 👀
Registration is now open: luma.com/bbh
Thanks to all of our partners, it's gonna be huge‼️
We're hiring!
At Gradium, we’re building the best voice models out there, with people we enjoy spending time with, in and out of the office.
Join us.
gradium.link/careers
We’re hiring across research, engineering, DevRel, sales, and marketing.
More details on open roles: gradium.link/careers
Gradium retweeted
Halloween is around the corner.
So I turned my Mac into a Haunted Mac.
Try to close it, and it'll scare you.
Slap it, and it'll remember what you did.
Pick it up, and it'll know.
You can ask your agent to create a scary voice with @GradiumAI or use one of the existing ones from the catalog.
Gradium and @covaldev are hosting an evening in Paris for people building voice agents.
Thursday, 5 Nov, 6 pm CET.
We’ll start with a short fireside chat between our research teams, followed by drinks and conversation with both teams.
RSVP: luma.com/wszf3b7v
Our CEO @neilzegh joined @Mattintouch on Génération Do It Yourself to discuss why voice will replace the keyboard as our way of working with AI agents.
youtu.be/oUo1Rneolgg?si=oOva…
We’re releasing a set of free tools powered by Phonon. The first is Gradium Read and Transcribe, a Chrome extension that reads any webpage out loud.
It runs on Phonon, our 40M-parameter multilingual model, and handles everyday text.
For hard cases like acronyms, phone numbers, and dates, you can switch to the Gradium API, which handles them by default.
gradium.link/read
Gradium now works hand in hand with Yampa: our TTS and STT now run natively in Yampa, orchestrated in real time by the Yampa Voice Engine. Customer care voice agents get low latency, robust pronunciation and custom voices, in SaaS on EU infra or on-prem. Full post: gradium.ai/blog/yampa-gradiu…
Gradium retweeted
I really like how snappy the voice is on my local @M5Stack StackChan Robot is. It's actually running @NousResearch hermes-agent and the voice is being handled by @GradiumAI Phonon. Parakeet handles TTS. I can run local models so everything is free, or I can use subscription based models as well. Hermes is model agnostic so that means you can choose what you want to use
Last year this robot was limited in its ability. Now it's a fully functioning intelligent thinking robot. Everyone has the ability to own their own today. This is peak awesomeness.
Built with grok @bot and hermes-agent. Team work
Our launch partners at @autonomous_labs gave Lamp, their personal robot with an open-source OS, a voice that it generates on the robot itself.
Lamp runs on an OrangePi 4 Pro, and Phonon shares its CPU with the robot’s OS and local AI models.
The team designed Lamp’s wizard-hat character voice in Gradium Voice Designer.