Nicolas Debock (@ndebock) header
Nicolas Debock (@ndebock) profile picture

VC @eurazeo (ex @balderton / @xangevc / @laposte). Tweet about the #web #tech #startups #mathematics . Fan of Claude Shannon. #FOCS

Paris, France
Joined February 2008
Photo by Nicolas Debock (@ndebock)Photo by Nicolas Debock (@ndebock)Photo by Nicolas Debock (@ndebock)Photo by Nicolas Debock (@ndebock)Photo by Nicolas Debock (@ndebock)Photo by Nicolas Debock (@ndebock)Photo by Nicolas Debock (@ndebock)Photo by Nicolas Debock (@ndebock)Photo by Nicolas Debock (@ndebock)Photo by Nicolas Debock (@ndebock)
Nicolas Debock retweeted
Just one week in, and our new open-source TTS model Phonon has already been downloaded ~85k times! So exciting to bring models to the open-source community. And check out our inference codebase in 🦀, it should make the most out of your local CPU/GPU/NPU/...
80K+ downloads and counting for Phonon. Here are @neilzegh and @lmazare answering some questions from the community. gradium.link/phonon
2
6
28
1,949
Nicolas Debock retweeted
Join us in Paris next week, together with our friends at @covaldev @bnicholehopkins
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
2
5
189
Bringing voice to any device. With models that are getting smaller. and it's open-source.
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🧵
2
1
8
241
Nicolas Debock retweeted
I fact checked this and it is depressingly accurate.
Almost the entirety of the British Highstreet is now made up of 3 types of businesses: 1) Private Equity owned companies (Pizza Express, Burger King, Cafe Rouge, The Body Shop, Wagamamas, Zizzi, New Look, Bella Italia, Boots, WH Smith, etc) 2) Foreign owned businesses acting as fronts for skilled visa migrant scams and/or money laundering enterprises (Hairdressers, Kebab Shops, Vape Shops, etc) 3) Charity Shops. Largely in existence to sell off items left when people die. It's difficult to imagine anything more dystopian than this.
8
9
119
184,421
Nicolas Debock retweeted
Fastest model out there.
Gradium-TTS is #1 on the Voice Arena US English 🇺🇸 latency board: 68 ms to first audio at the median, the fastest model we have measured. Gradium-TTS is the latest TTS model from @GradiumAI. It comes in ahead of SpeechifyAI's Simba 3.2 at 123 ms and Inworld TTS 2 at 168.5 ms, and more than three times faster than its own earlier 236 ms. Speed without giving up quality: on the same board it ranks #10 of 21 at 1007 Elo from blind native-speaker votes, which puts it on the quality-versus-speed frontier with Simba 3.2, Inworld TTS 2 and Cartesia's Sonic-3.6. We measure time to first audio the way a listener hears it, with leading silence counted and fresh text on every request so nothing is served from cache. Congratulations to @GradiumAI on the release! See below for the quality board, what raters said, and the exact audio they judged 🧵
1
7
49
4,595
Nicolas Debock retweeted
Hear me out, 💻120 000 B300s 🗼Tour Montparnasse. 🔌304MW
19
35
12
478
52,712
Nicolas Debock retweeted
Our first build with Jev from @typesafeai: a support agent that design its voice from the call context. What would you build? gradium.ai/blog/voice-that-a…
1
3
2
17
962
Nicolas Debock retweeted
We just released our latest TTS model, time to first audio went from ~250ms to ~50ms 🚀. Co-designing the model architecture and a custom inference stack is what made such an improvement possible. If you want to make some GPUs go brrrr running some audio models, we're hiring!
Gradium TTS beta model is out now to define what fast means. Less than 50ms TTFA, same quality. Most models are >100ms TTFA. Try it now: studio.gradium.ai/?model_nam…
1
4
29
2,407
Nicolas Debock retweeted
100 years old. Still the sharpest mind in the room. Jean-Pierre Serre, youngest Fields Medalist in history (age 27), first Abel Prize winner, and the “Mozart of mathematics,” turned 100 on September 15 and is still lecturing, still writing on the blackboard, still shaping the field. Happy 100th to the greatest living mathematician. The rest of us are just trying to keep up.
58
883
75
5,197
310,325
Nicolas Debock retweeted
In this thread is a link to an AI creativity engine I've started working on. At university I switched from engineering to architecture and had to unlearn being over scientific and learn creativity.
1
1
5
762
Nicolas Debock retweeted
we've sandboxed the agent the agent:
57
711
61
7,735
212,500
Nicolas Debock retweeted
18x improvement in intelligence per joule in 16 months.
hard agree with @amasad —@JonSaadFalcon and my research indicates that intelligence efficiency (intelligence per watt) is rapidly improving and we will definitely *not* need data center scale compute to run agi! links to research in comments below 👇
146
223
99
1,709
6,023,620
Nicolas Debock retweeted
Sad none of my "joke" proposals made the final cut.
Ten selected design proposals for the next series of euro banknotes have been unveiled. Explore the proposals and tell us which ones you prefer. Have your say surveys.ecb.europa.eu/10b/ec…
169
4,156
178
75,428
3,688,989
This fantastic figure by @jburnmurdoch is Exhibit 1 of what an aging society means for the political game: public investment, which is choosing future rewards over present consumption, gets squeezed out. Still not convinced this is a first-order challenge?
45
492
62
1,746
213,556
Nicolas Debock retweeted
Pas besoin de technologie compliquée pour atteindre l'invisibilité : 25 rotations / seconde suffissent. L'université Northwestern 🇺🇸 présente un drone rotatif qui se dissimule, en exploitant le « flou de mouvement » et une faille de la vision humaine. news.northwestern.edu/storie…
27
105
8
541
78,752
Nicolas Debock retweeted
The Google Brain (& @GoogleDeepMind) team back in 2015 feels like a stronger fit to this picture: Andrej Karpathy, Ilya Sutskever, Dario Amodei, Noam Shazeer, Demis Hassabis, Shane Legg, Mustafa Suleyman, Jeff Dean, Andrew Ng, Greg Corrado, Geoffrey Hinton, Ashish Vaswani, Jakob Uszkoreit, Łukasz Kaiser, Niki Parmar, Aidan Gomez.
28
107
9
1,974
182,877
Nicolas Debock retweeted
Yep, we did.
This Post is from a suspended account. Learn more
24
28
6
560
244,809
Nicolas Debock retweeted
Theory of relativity in a single picture
48
2,298
44
36,828
391,513
Nicolas Debock retweeted
In 1948, a 32-year-old at Bell Labs published a paper nobody fully understood. Engineers found it too mathematical. Mathematicians found it too engineering-focused. One prominent mathematician reviewed it negatively. That paper - "A Mathematical Theory of Communication", became the founding document of the digital age. The man was Claude Shannon. Father of Information Theory. At 21, he wrote the most important master's thesis of the 20th century. Working at MIT on an early mechanical computer, Shannon noticed its relay switches had exactly two states - open or closed. He had just taken a philosophy course introducing Boolean algebra, which also operated on two values: true and false. Nobody had ever connected these two things. His 1937 thesis proved that Boolean algebra and electrical circuits are mathematically identical, and that any logical operation could be built from simple switches. Howard Gardner called it "possibly the most important, and also the most famous, master's thesis of the century." Every digital computer ever built traces back to this insight. At 29, he proved that perfect encryption exists. During WWII, Shannon worked on classified cryptography at Bell Labs. His work contributed to SIGSALY, the secure voice system used for confidential communications between Roosevelt and Churchill. In a classified 1945 memorandum, he mathematically proved the one-time pad provides perfect secrecy, unbreakable not just computationally, but provably, permanently, against an adversary with infinite power. When declassified in 1949, it transformed cryptography from an art into a science. It laid the foundations for DES, AES, and every modern encryption standard. At 32, he defined what information is. His 1948 paper introduced one equation: H = −Σ p(x) log p(x) Shannon entropy. The average uncertainty in a probability distribution. The minimum bits required to encode a message. Three things followed: > He defined the bit - the fundamental unit of all information. His colleague John Tukey coined the name. > He proved the channel capacity theorem, every communication channel has a maximum rate of reliable transmission. You can approach it. You can never exceed it. > He unified telegraph, telephone, and radio into a single mathematical framework for the first time. Robert Lucky of Bell Labs called it the greatest work "in the annals of technological thought." Where his equation lives in AI today: Cross-entropy loss - the function training every classifier and language model, is derived directly from H. Decision tree splits use information gain, which is H applied to data. Perplexity, the standard LLM evaluation metric, is an exponentiation of cross-entropy. Every time a neural network trains, Shannon's formula runs inside it. He also built the first AI learning device. In 1950, Shannon built Theseus, a mechanical mouse that navigated a maze through trial and error, learned the correct path, and repeated it perfectly. Mazin Gilbert of Bell Labs said: "Theseus inspired the whole field of AI." That same year he published the first paper on programming a computer to play chess. He co-organized the 1956 Dartmouth Workshop, the founding event of AI as a field. The man: He rode a unicycle through Bell Labs hallways while juggling. He built a flame-throwing trumpet, a rocket-powered Frisbee, and Styrofoam shoes to walk on the lake behind his house. He called his home Entropy House. When asked what motivated him: "I was motivated by curiosity. Never by the desire for financial gain. I just wondered how things were put together." In 1985, he appeared unexpectedly at a conference in Brighton. The crowd mobbed him for autographs. Persuaded to speak at the banquet, he talked briefly, then pulled three balls from his pockets and juggled instead. One engineer said: "It was as if Newton had showed up at a physics conference." He died in 2001 after a decade with Alzheimer's, the cruel irony of information slowly leaving the mind of the man who defined what information was. Claude, the AI model, is named after Claude Shannon, the mathematician who laid the foundation for the digital world we rely on today.
195
2,070
217
7,577
467,697