Deepgram

Independent benchmarks just confirmed it: Aura-2 leads on real-time TTS latency. Coval recently added Aura-2 to their public TTS benchmarks. They test across real-world scenarios, measuring latency, consistency, and cost under production conditions. The results: ⚡ Lowest latency: Aura-2 delivers the fastest time to first byte among models tested. Median latency under 90ms, with 95th percentile under 200ms. 🧠 Tightest distribution: Low averages matter, but so do long-tail spikes. Aura-2 shows minimal variability. Fewer awkward pauses, more predictable behavior for SLAs. 🚀 Cost efficient at scale: Aura-2 sits in the lower-left quadrant: fast responses + competitive pricing. Few models combine speed, consistency, and cost efficiency in the same region. For production voice agents handling thousands of calls daily, this matters. A 100ms difference per response compounds into hours of reduced wait time. Behind the scenes: Our team cut TTFB from sub-200ms at launch to ~90ms today through Rust-based runtime optimization, improved GPU orchestration, and tighter scheduling. Coval’s benchmark explorer is public. You can examine Aura-2’s performance directly and compare against other models. Full breakdown in the links below: - Read the full breakdown here: https://lnkd.in/gSyVm-rv - Explore Coval’s benchmarks: https://lnkd.in/g7dSREmr - Try Aura-2 in the Deepgram Playground: https://lnkd.in/gJQyEkiw

AI voice agents live or die in the first few seconds of a call. If the voice is slow or robotic, people hang up. Jobcase, Inc., who supports millions of frontline and hourly workers, tested multiple TTS providers (AWS Polly, OpenAI, ElevenLabs) and chose Deepgram Aura-2 TTS for the best mix of quality, cost, and latency. Now their AI-driven calls: ➡️ Start speaking faster ➡️ Sound more natural ➡️ Scale to higher concurrency without breaking the experience If you’re still treating TTS as a commodity, you’re leaving user experience (and throughput) on the table. Read how Jobcase powers AI voice calls with Aura-2 TTS: https://lnkd.in/eHytFKrD

We couldn’t agree more. As AI interaction moves from text to voice, the infrastructure layer matters—and we’re proud to be building it alongside partners like @AVP who deeply understand platform shifts. Onward. 🔊⚡

🚀 Thank you to Alex Edwards at Slator for the tremendously insightful discussion with our CEO Scott Stephenson. We're so glad to show the world not only the technology we have available now, but also the AI advancements we plan to offer in the future. Link: https://lnkd.in/gxVGZQ3B

Your LLM provider went down at 2am. Your voice agent went with it. We kept hearing this from teams running production voice agents. One vendor hiccup, and suddenly you’re scrambling to reconfigure agents while calls fail. So we built LLM fallbacks into Voice Agent API. Here’s how it works: → Configure multiple LLM providers for your agent’s “think” step → If your primary returns an error, Voice Agent API automatically fails over to the next → No manual intervention. No dropped calls. We also added support for the latest models: • OpenAI GPT 5.1 and GPT 5.1 Chat • Google Gemini 3 • Anthropic Claude Sonnet 4.5 Use them as your primary, your fallback, or both. This isn’t smart routing or load balancing; it’s ordered failover when things break. Production resilience, built into the runtime. Docs link in comments. What’s your current fallback strategy for LLM outages?

Huge shoutout to Built In San Francisco for showcasing not only our latest round of funding, but also what our technology can do! We’re so proud to provide low-latency, real-time Voice AI to developers worldwide Link to Article: https://lnkd.in/gKWn8a8a

We’re absolutely honored to have hour latest funding round featured in the The Wall Street Journal VC Pro Private Equity Newsletter! Thanks so much to for spreading the news of our latest milestone 🚀 Link to feature: https://lnkd.in/gbMcKEag

We’ve had a big week—announcing our Series C, collaborating with our partners to build the future, and thanking our investors for their continued confidence in our ability to deliver. However, all of these incredible updates are only possible because of one key factor: our team. Whether you're in sales, engineering, research, or HR, we'd love to work with you if you are: 🌍 Collaborative: DGers are generous and help each other. We value teamwork and the contributions of each individual. 🔎 Curious: We continually improve our software and work methods, challenging the status quo and setting new boundaries. 🤝 Customer-First: We listen to our customers to grasp their goals, making our innovation strong by adapting to their needs. ☮️  Always trying to do the right thing: We are kind, assume the best intentions, and have implicit trust in each other and our customers. At Deepgram, we grow together, think smart, and act fast. If that sounds like you, come take a look at our open roles. We’d love to meet you. 👉 Apply here: https://lnkd.in/gDkZ2f4E

Thank you, Kore.ai! We are thrilled to partner with you as you help global brands build, deploy, and orchestrate AI agents across the enterprise. 🔊

After all the news about our Series C and acquisition of OfOne, some of our new followers might wonder why so many companies trust us to be the backbone of their Voice AI infrastructure. Why is being “Powered by Deepgram” such a game-changing innovation? Well if you’re curious how our models actually perform, don’t just take our word for it. You can try them yourself right now! 👉 🧪 Test our voice AI models in real time—speech-to-text, text-to-speech, and voice AI agents: https://lnkd.in/gn8gUHBF Tag: OfOne (acquired by Deepgram)