Real-time APIs powering the Voice AI economy. Trusted by enterprises & 200k+ developers to build low-latency Voice AI.
San Francisco, CA
Based in United States
- Tweets 2,799
- Following 745
- Followers 11,262
- Likes 2,343
Voice AI is growing because builders keep pushing it forward, and we believe in backing teams early.
On top of the Deepgram Startup Program, we're proud to announce that we're a launch partner in @livekit's Startup Program. Pre-Series A teams get $1,000 in Deepgram credits and it works natively through LiveKit Inference.
Ready from your first line of code and as you scale.
Can't wait to see what you build: livekit.com/startups
2
1
5
379
Deepgram is ranked #17 on RepVue's Best Venture Backed Companies to Sell For! 🏆
We're especially proud because this isn't an award you apply for. Rather, rankings come from anonymous feedback by people who actually sell here.
We're hiring AEs. 📈
repvue.com/rankings/venture-…
1
1
3
407
Some of the biggest Voice AI opportunities live in places where the cloud isn’t always an option.
That’s why Deepgram is excited to support @VAST_Data's launch of DataEnclave – bringing leading AI models into customer-controlled infrastructure while helping protect sensitive data and models in use.
Deepgram already supports self-hosted Voice AI for enterprises that need greater control over where their workloads run. With VAST DataEnclave, those teams have another path to put real-time Voice AI to work in the environments that matter most.
Because the best AI infrastructure meets enterprises where they are.
More on DataEnclave: bit.ly/4hhYt4T
6
2
12
561
Introducing Nova-3 Pharma, the first speech-to-text model purpose-built for the pharmaceutical industry.
It’s optimized for drug names, pharmaceutical terminology, and medical vocabulary. 🧵👇
3
5
14
915
The Deepgram India endpoint is now generally available. api.in.deepgram.com runs in Hyderabad, with storage and inference for your audio, transcripts, and speech output in India.
Same API and models. Moving over is a base URL change, no self-hosting required.
Blog → deepgram.com/learn/deepgram-…
1
1
6
501
Speak '26 registration is officially open! 🔥
Voice AI is on the verge of its biggest milestone yet. On Oct 29, the people building it are getting together in SF to talk about what it takes to get there.
Register: deepgram.com/speak
3
2
11
1,520
Deepgram has a new Chief Product Officer: welcome, Ed Anuff. 👋
Ed has advised Deepgram since 2021. Now he’s joining us to lead product strategy, staying close to how customers are building with voice AI and helping shape what comes next.
Big things ahead. 🤝
2
1
13
925
The voices with good taste. 🤌
Flux TTS handles menu pronunciation, prices, hours, natural turn-taking, and conversational timing.
Plus, you can pick a voice that sounds (and feels) like your brand.
Try the restaurants voice AI ordering demo: deepgram.com/solutions/voice…
5
586
Our customers said it best. We gave Flux the mic. 🎤
Hear from @DoccyHealth, @trydaily/@pipecat_ai, and @livekit on taking TTS beyond line-by-line speech and into real-time, conversation-aware voice experiences.
Explore Flux TTS: deepgram.com/product/text-to…
3
6
13
1,433
Consider this your sign to chat with Flux TTS.
Discover what you can build: deepgram.com/product/text-to…
4
1
13
865
Nova-3 just got even better across 50 languages.
We’ve rolled out monolingual quality upgrades across 50 languages, improving transcription accuracy and keyterm prompting across batch and streaming STT, plus added 8 new languages: Afrikaans, Armenian, Georgian, Nepali, Punjabi, Assamese, Pashto, and Mongolian.
5
2
27
1,927
Deepgram is a @G2dotcom Leader in the Fall 2026 Grid Reports across voice AI, including Text-to-Speech.
Deepgram also claimed the #1 spot in Voice Recognition and is the only one of G2’s five Voice Recognition Leaders to also earn a Leader position in the Text-to-Speech G2 Grid.
That momentum in TTS continues with Flux TTS, built to make real-time voice conversations sound and feel more natural.
7
979
Clear your calendar for October 29th.
Deepgram Speak ’26 is landing in San Francisco.
Save the date → deepgram.com/speak
5
2
14
1,587
There’s no “close enough” button in compliance.
2X Solutions AI uses Deepgram’s Nova-2 to catch opt-outs in real time across 7.3M+ calls—with sub-2-second transcription turnaround.
When the transcript is the evidence, every word counts. Full case study: deepgram.com/customers/2x-so…
7
8
554
Your voice agent can remember the conversation. Its voice should sound like it does, too.
That's what Flux TTS is built for.
Instead of starting fresh every time the agent speaks, Flux TTS carries conversational context across turns, helping tone, pacing, and pronunciation stay consistent without SSML or style tags.
And yes, it works with @pipecat_ai.
11
1
45
2,985
Imagine writing this before every line: 🙃
[empathetic] “I understand how frustrating that must be.”
That’s the work Flux TTS removes.It reads the conversation and uses what came before to determine how the next line should be delivered.
No SSML. No style tags. No extra prompt engineering.
talk.deepgram.com?utm_source…
19
21
1,745
Enterprise voice AI has to work on real calls, not just clean demos.
GetVocal uses Deepgram for fast, accurate STT on telephony audio and delivers 70% deflection, 45% more self-service resolutions + 31% fewer live escalations.
Full case study: deepgram.com/customers/getvo…
2
6
506
Voice AI is moving from demos to operations.
Omnico runs customer interactions AI-first across voice, chat, browser, and WhatsApp. Deepgram powers the speech layer built for real-world conversations at enterprise scale.
Put the future of #CX to work today.
2
3
13
1,160
Tomorrow, voice AI meets its toughest test: a real phone call. 📞
We’re teaming up with @twilio to break down what it actually takes to ship voice AI in production.
⏰ 10 AM PT. Save your spot → deepgram.com/webinars/voice-…
1
7
617
Singapore, we're moving in. 🇸🇬
We’re establishing our APAC headquarters in Singapore, supported by a strategic investment from EDBI.
Why now? Deepgram’s API traffic in APAC increased 96% YoY, and we're serving customers in over 20 markets. deepgram.com/learn/voice-ai-…
1
5
556
Deepgram CEO @deepgramscott joined @furrier from @theCUBE for #NYSEWired to talk Flux TTS, Deepgram crossing $100M ARR, and voice as a primary interface.
Catch the whole conversation: piped.video/watch?v=D3xDOYbE…
5
17
1,050
Summer’s heating up. A new lineup of voices just entered the villa. 🔥
Meet Flux TTS: expressive voices built for live conversation.
And these four? They’ve got good chat.
Ready to meet your match? Try Flux TTS today: talk.deepgram.com?utm_source…
2
15
1,152
Today, we're launching Flux TTS: expressive voices built for live conversation.
We asked our very own Lauren O’Connor to put Flux TTS to the test.
Most TTS handles one line at a time. Flux TTS carries context across the whole conversation. And it's a conversation you'll actually want to have.
Hear how Flux meets each moment.
22
15
156
12,113
We got a text. 📲
New voices are entering the villa, and they just might turn some heads.
August 12. 👀
1
2
8
700
🔐 Self-hosted AI shouldn’t mean slower support.
With Deepgram on Amazon SageMaker AI:
• Models stay in your AWS environment
• Support stays fast and secure
• Initial ticket investigation moves from days to minutes
Our very own Victor Wang teamed up with @AWSCloud to share how it works.
aws.amazon.com/blogs/machine…
3
5
608
Turning it on is a config change
Same API, same deploy, same Nova models. No difference in performance vs non-FIPS.
1
1
2
71
For regulated buyers, the speech AI eval often ends on one question:
"Is the cryptography FIPS validated?"
If the answer is no, the on-prem deployment is dead. No matter how good the models are.
We just changed the answer. 🧵
2
2
5
626
Our first London Voice AI meet-up kept circling one question: how much of human conversation does a machine really need to understand? 🧐
Our take: the deeper the understanding, the better the orchestration. Get it right and customers ship their product, not voice plumbing.
1
2
7
560
Voice AI, minus the cloud round trip.
We're bringing Deepgram's Nova-3 on-device, optimized for the @Qualcomm Hexagon NPU in Snapdragon X Series PCs. Real-time speech recognition with more speed, more privacy, no connection required.
deepgram.com/learn/deepgram-…
2
6
600
We took voice AI to Madrid. The room was ready. 🇪🇸
At ExpoContact, Miguel Aroja discussed what it takes to go from pilot to production in the contact center. Latency below 200ms. 50+ languages. Real enterprise proof.
Contact centers are ready for voice AI. Are you?
1
3
569
New release! Flux word-level timestamps are here. 🎙️
Every word. Exactly when it was spoken. developers.deepgram.com/chan…
1
3
8
644
Your party deserves a historian. 📜
Jabari Barton built CritScribbler so D&D players never lose the plot. Session audio goes in, searchable notes and campaign lore come out, built by Deepgram.
We love when our community builds the tool they wished existed ⚔️🎲
2
653
🎙️✨ @Claude Sonnet 5 is now available in the Voice Agent API. 🎙️
Improved reasoning. Better conversational quality. Ready for your voice agents today.
Set the model in your agent configuration and start building. Full details and supported models in the changelog. developers.deepgram.com/chan…
2
2
533
🍔👑 Here's what @QSRmagazine cover story on @BurgerKing tells us about voice AI.
Tens of thousands of guest calls. Feedback that changed products and campaigns. 10th to 6th in guest experience.
None of that works without a speech engine that understands people.
Counter. Kiosk. App. Voice is now.
That engine is what we build.🎙️ issuu.com/wtwhmedia/docs/qsr…
3
1
1
402
Aura-2 is built for real-time text-to-speech: fast, accurate, and cost-efficient at scale.
🎧 Try it in the Deepgram Playground: playground.deepgram.com/
2
2
8
839
First Developer Spotlight is live! And we're opening with one of our favorites.
CATT (@itscatt_yt): a real-time transcription and translation tool for VTubers from Japan and Korea streaming to Western and SEA audiences, built with Deepgram.
🧵👇
2
1
2
1,819
Great connecting with customers, partners, and builders at #CCWVegas. 🎰
Thanks to everyone who joined Deepgram, @awscloud, and @koredotai for an evening in the @Raiders Locker Room!
See you at the next one. 👋
2
1
514
Voice AI is moving from feature to foundation.
Our latest research highlights seven shifts in how organizations are thinking about adoption, investment, and voice agents.
Here are the findings that stood out.🧵
3
3
7
559
Data + AI Summit is wrapping up. But the Voice Heist is still live! Think you can break our voice agent? 🏆
The challenge sparked some of the best conversations we've had about deploying voice AI in production.
Come see us at booth #124 to take on the challenge. 🦸 #DataAISummit
2
1
1
350
The Deepgram Australia endpoint is now generally available. 🇦🇺
api.au.deepgram.com runs in Sydney, with your audio, transcripts, and speech output processed and stored entirely within Australia.
Same API and models. Moving over is a base URL change, no self-hosting required.
Blog → deepgram.com/learn/deepgram-…
1
2
301
Day 1 at #HLTHEurope, and one thing is clear: healthcare teams aren't asking if they'll use voice AI anymore. 🩺
We're seeing strong interest in how teams move from experimentation to real-world deployment, particularly when it comes to self-hosted deployments, EU endpoints, and meeting regulatory requirements.
Looking forward to more conversations and insights throughout the week. 🚀
1
1
1
273
Introducing Batch Diarization V2, a major upgrade to speaker labeling for pre-recorded audio.
Understanding what was said is only part of the story. You also need to know who said it.
Highlights:
🔹 Preferred 3.3X in human evaluation
🔹 Improved speaker attribution accuracy across real-world audio
🔹 Available today via the new diarize_model parameter
Get started: deepgram.com/learn/introduci…
1
3
300
See how @premai_io brought real-time voice to its sovereign AI ecosystem with Deepgram Nova-3, delivering low-latency, multilingual speech-to-text that never leaves trusted infrastructure. 🔒
Read the full customer story: deepgram.com/customers/prem-…
2
1
7
455
We're heading to #NiCEWorld26 next week. 🚀
Looking forward to connecting with customers, partners, and innovators shaping the future of #CX. See you there. 👋
1
2
310
The improved Nova-3 Portuguese model is here! 🇵🇹 🇧🇷
We’ve enhanced the Nova-3 Portuguese model with improved transcription accuracy across Portuguese language variants, including Brazilian Portuguese and European Portuguese.
Available now with: model="nova-3"
Supported language codes: pt, pt-BR, pt-PT
Try it in the Playground or get started in the docs:
playground.deepgram.com/?end…
1
6
747