Cartesia

Today, we’re unveiling a significant milestone in our journey toward ubiquitous artificial intelligence: On-Device. Our team pioneered a radically more efficient architecture for AI with state space models (SSMs). Now, we’ve optimized and deployed them at the edge. We believe the future of AI runs on your device, where it can process continuously and is reliable, private and secure. Read our full blog post here: cartesia.ai/blog/on-device As part of this milestone, we’re announcing Edge, Rene, and Sonic On-Device. 📏 Edge: an open-source library for on-device applications across all environments, starting with Apple hardware. 🐍 Rene: an open-source 1.3B parameter language model with a hybrid Mamba-2 SSM backbone, tailored to on-device inference. Now on Edge. 📲 Sonic On-Device: The Sonic model you know and love, running on-device in private beta. It’s the first ultra-realistic voice model of its kind to support real-time streaming on devices. We’re excited to keep building for the next generation of hardware-optimized use cases, allowing any device to benefit from real-time, multimodal AI. If you're interested in partnering with us on any of our On-Device initiatives, please reach out via our form at bit.ly/cartesiaondevice.

🗝 We’re excited to announce that Cartesia is now SOC 2 Type 2 Certified with no exceptions! 🗝 We are committed to safeguarding our users’ data and privacy, and this certification is another step towards ensuring that we provide secure, reliable and trustworthy services to all our customers. 🔐

We're thrilled to welcome Ivory Tang our team, who will be focused on growth strategies and developer engagement. 🚀 Ivory brings a unique blend of AI expertise and business acumen to our organization. As the founder of ScaleConvo (YCW24), a customer support AI agent, she has hands-on experience with leading voice providers in the industry. Her impressive background includes roles in consulting at BCG and technical stints at Jane Street and HubSpot. Ivory graduated from MIT where she studied Mathematics, Computer Science, and Business. During her time there, she also conducted research at the Computer Science and Artificial Intelligence Laboratory (CSAIL). On joining our team, Ivory shared: "Having spent over a year in the voice AI agent space, I've tried every existing provider. Cartesia stood out as the only one that successfully blended a research-focused approach with top-tier voice quality and low latency." We're excited about the growth initiatives Ivory will be spearheading. Stay tuned for updates 👀

🎉 Congrats to the team at toby for earning a top spot on Product Hunt with their recent launch! 🎉 We’re excited to share how we’re supporting them on their mission to empower global workforces with live speech translation on every video call. Toby chose to launch on Sonic because of our: 🚀 Industry-leading latency: Sub-100ms across ALL languages 💬 Multilingual Support: Consistent quality and latency across major languages 👄 Localization Control:  Match voices to local accents Read more of their story here 👇

Aviv Bick, one of Professor Albert Gu's students and a summer intern here at Cartesia, co-authored “Transformers to SSMs: Distilling Quadratic Knowledge to Subquadratic Models,” outlining a new way to distill quadratic Transformers into more efficient, subquadratic architectures like Mamba. By leveraging pretrained Transformers, they achieved state-of-the-art Mamba models using 1% of the compute! The key insight here was to project the (quadratic) attention matrices onto (structured) SSM matrix mixers before end-to-end training. Congrats to Aviv and his co-authors on this new publication! Blog ✍: https://lnkd.in/gwmRpDdG Paper 📜 : https://lnkd.in/gESrWSei

Congrats to the team Daily on launching Daily Bots 🎉 We're thankful to get to work with them as their default voice provider. Developers love Pipecat for its: 🏃♂️ ultra-low latency—perfect for real-time applications 💡 rich example library that enables the creation of AI agents in minutes 🚿 clean abstractions Plus, with the recent release of the open standard for Real-time Voice and Video Inference (RTVI-AI), Pipecat continues to set the pace in the industry. We're lucky to get to work with Kwindla Hultman Kramer and his team in pushing the boundaries of multimodal AI.

We’re excited to welcome Brandon Wang to our technical staff 🎉 . Brandon recently graduated from MIT where he studied computer science, molecular biology, and math. While at MIT, Brandon was an active researcher in the Computer Science and Artificial Intelligence Library (CSAIL), and published a paper on Constant-Time Routability that was featured in the 2024 Symposium on Theory of Computing. Brandon also previously interned at Jane Street and was a 2019 International Math Olympiad Gold Medalist. “I heard about Cartesia through a college friend who loved his intern experience here. I’m excited to work on challenges like making AI more accessible with a really great group of people,” he said. You can read Brandon's most recent paper below 👇

Excited to discuss all things real time AI with our friends Pilot.com!

Super excited to partner with Tavus and power the fastest conversational video experience with Sonic. We believe the next generation of real-time, multimodal AI experiences are going to change everything. Try talking to Carter, it's responsive and super fast!

We are live on AudioStack! We've had a blast working with Peadar Coyle and this brilliant team to build the future of AI-powered production. Use one of our 94 ultra-realistic voices across 7 languages (🇺🇸🇪🇸🇫🇷🇩🇪🇨🇳🇯🇵🇧🇷) to bring life to your text across videos, ads and more!