Latest News & Updates on Deepgram

Deepgram has launched Flux TTS, a conversation-aware text-to-speech model for real-time voice agents. Flux TTS maintains conversational context, ensuring consistent tone and pacing across interaction...s. It supports various deployment options, including cloud and on-premises, and integrates with platforms like Vapi and Cloudflare. The model is available globally, with standard pricing starting September 13, 2026.
Show More..

Deepgram has launched Flux TTS, a conversation-native text-to-speech model designed for enterprise voice agents. Flux TTS integrates with Deepgram's Voice AI platform, offering low-latency, context-a...ware speech generation. It supports enterprise needs with deployment flexibility and is now generally available, with a free trial until September 12, 2026. The model enhances voice agent performance in complex, real-world interactions.
Show More..

Deepgram has integrated its Nova-3 speech-to-text model with Qualcomm's Snapdragon processors, enabling real-time voice AI on devices without cloud dependency. This integration allows for enhanced vo...ice experiences across automotive, mobile, and IoT applications, addressing latency and privacy concerns.
Show More..

Deepgram has optimized its Nova-3 speech-to-text model for Qualcomm's Snapdragon X Series platform, enabling real-time speech recognition on PCs without cloud reliance. This integration enhances spee...d, privacy, and reliability, supporting diverse applications like automotive and IoT. Nova-3 offers multilingual transcription with a 24.7% lower error rate than competitors.
Show More..

Deepgram has integrated its Nova-3 speech-to-text model with Qualcomm's Snapdragon processors, enabling real-time voice AI on devices without cloud dependency. This integration enhances speed, privac...y, and reliability, supporting applications across automotive, mobile, and industrial sectors. Nova-3 offers multilingual transcription and superior accuracy with a 6.89% word error rate.
Show More..

Deepgram has optimized its Nova-3 speech-to-text model for Qualcomm's Snapdragon processors, enabling real-time voice AI on devices without cloud reliance. This integration enhances speed, privacy, a...nd reliability for applications across various sectors, including automotive and IoT. Nova-3 offers superior accuracy with a 6.89% word error rate and supports multilingual transcription and instant vocabulary adaptation.
Show More..

Deepgram optimizes its Nova-3 speech-to-text model for Qualcomm's Snapdragon X platform, enabling on-device speech recognition without cloud reliance. This integration enhances real-time voice applic...ations across sectors like automotive and IoT, offering faster, reliable experiences with a 6.89% word error rate.
Show More..

Deepgram and Qualcomm have partnered to integrate Deepgram's Nova-3 speech-to-text model with Snapdragon X Series processors, enabling on-device, real-time voice recognition without cloud dependency.... This integration targets AI PCs, automotive, and industrial edge deployments, offering a 6.89% word error rate and customizable domain-specific vocabulary. The partnership marks a shift towards edge-native architectures in the voice AI industry.
Show More..

Deepgram has launched its Australia endpoint, enabling customers with Australian data-residency requirements to process speech-to-text and text-to-speech workloads within the country. This fully mana...ged service, hosted in AWS's Sydney region, ensures compliance with local regulations, particularly benefiting healthcare, financial services, and government sectors. The endpoint supports the same APIs and models as Deepgram's Global and EU regions.
Show More..

Deepgram's voice AI platform now supports fully on-premises deployment with end-to-end encryption, powered by Fortanix and NVIDIA Confidential Computing. This integration targets regulated industries... like banking and healthcare, enabling compliance with data-residency rules by keeping sensitive audio data and AI model weights encrypted within hardware-protected environments.
Show More..

Deepgram has partnered with Fortanix and NVIDIA to offer secure, on-premises voice AI deployments for regulated industries. This integration uses Fortanix Confidential AI and NVIDIA Confidential Comp...uting to protect sensitive data and AI model weights, enabling enterprises to meet stringent security and regulatory requirements while maintaining real-time performance.
Show More..

Deepgram's Nova-3 expands its speech-to-text capabilities across Asia-Pacific, adding support for Thai, Cantonese Traditional, Mandarin Simplified, and Mandarin Traditional. This update enhances tran...scription accuracy, achieving significant Word Error Rate reductions compared to Nova-2. Nova-3 also improves recognition for Indic languages like Bengali and Marathi, supporting diverse speech environments. These enhancements cater to developers and enterprises, offering seamless integration via API for both streaming and batch workflows.
Show More..

Deepgram has launched three new agentic engineering tools for voice AI developers: the dg CLI, MCP server, and deepgram/skills repo. These tools streamline the development process by integrating with... AI coding tools, enabling seamless transcription, speech generation, and API management. The platform supports multiple programming languages, enhancing productivity and reducing development time.
Show More..

Deepgram has launched Flux Multilingual, a conversational speech recognition model supporting 10 languages, enabling seamless language switching and accurate dialogue flow. It offers features like mo...del-based turn detection and compatibility with existing Flux API integrations. Available via Deepgrams Cloud API or self-hosted, it aims to enhance global enterprise customer interactions through voice agents.
Show More..

Deepgram has launched Flux Multilingual, a speech recognition model supporting ten languages with mid-call language switching capabilities. The model, available via Deepgram's cloud API, offers low-l...atency turn detection and interruption handling, enhancing integration for real-time voice agents. It supports languages like English, Spanish, and Hindi, and includes a language_hint parameter for biasing recognition.
Show More..

Deepgram has launched Flux Multilingual, enhancing its conversational speech recognition platform to support real-time multilingual understanding across ten languages. This innovation integrates tran...scription with conversational awareness, allowing AI agents to handle natural language dynamics like turn-taking and interruptions. Flux simplifies deployment by using a single model for multiple languages, reducing complexity and improving interaction quality for enterprises in sectors like healthcare and finance.
Show More..

Deepgram has launched Flux Multilingual, a conversational speech recognition model supporting 10 languages. This model enables real-time, multilingual interactions with features like native turn dete...ction, interruption handling, and code-switching. It simplifies building global voice agents without complex integrations, offering monolingual-grade accuracy and low latency. Flux Multilingual is available via Deepgram's Cloud API or as a self-hosted deployment.
Show More..

Deepgram has released "deepclaw," an open-source integration that enables users to call their OpenClaw AI assistant via phone using Deepgram's Voice Agent API. This integration leverages Deepgram's F...lux speech-to-text and Aura-2 text-to-speech technologies, offering seamless turn-taking and reduced latency. The setup is straightforward, requiring minimal technical expertise, and the solution is fully open source, allowing for customization and self-hosting.
Show More..

Deepgram has launched a new Arabic speech-to-text model on Nova-3, covering 17 regional dialects. It offers up to 40% lower word error rates compared to competitors. The model supports cloud API and ...self-hosted deployment, enhancing flexibility for developers and enterprises. This release strengthens Nova-3's position in global voice AI, particularly for Arabic-speaking regions.
Show More..

Deepgram has launched a native integration with Amazon SageMaker AI, offering real-time speech-to-text, text-to-speech, and voice agent capabilities with sub-second latency. This integration allows e...nterprises to deploy voice-powered applications within AWS environments, enhancing security and compliance. Deepgram's collaboration with AWS includes a multi-year Strategic Collaboration Agreement to boost enterprise adoption of voice AI technologies.
Show More..

Deepgram has developed a groundbreaking speech-to-speech (STS) model that eliminates the need for text conversion, enhancing real-time voice interactions by preserving nuances and emotional tones. Th...is innovation, part of Deepgram's enterprise-ready voice AI platform, promises improved latency and naturalness, offering businesses a competitive edge in deploying scalable voice AI solutions.
Show More..

Deepgram has launched Nova-3, its latest AI-based speech-to-text model, enhancing transcription accuracy in challenging environments like noisy settings and specialized fields such as medical and leg...al. Nova-3 supports real-time multilingual transcription, self-service customization, and accurate numeric recognition, offering enterprises improved compliance and data privacy. The model's advanced architecture and extensive training cater to diverse industry needs.
Show More..

Deepgram has launched an improved version of its Flux transcription model, Flux V0.1, using reinforcement learning. This update enhances transcription accuracy by up to 10% for certain audio types an...d reduces false positives in start-of-turn detection by 70%. The model also offers faster end-of-turn detection, improving latency by 50-150ms. These enhancements aim to optimize end-of-turn accuracy without requiring developers to make changes.
Show More..

Deepgram's Aura-2 text-to-speech (TTS) engine now supports Dutch, French, German, Italian, and Japanese, expanding its multilingual capabilities. This enhancement allows developers to deliver consist...ent, natural voice experiences for global applications. Aura-2 excels in business use cases, offering clarity and low-latency performance, crucial for real-time voice agents. The update reflects Deepgram's commitment to enterprise-grade voice AI.
Show More..

Deepgram has announced the general availability of its EU Endpoint, enabling voice AI workloads to be processed entirely within the European Union. This release supports full EU data residency, lower... latency, and access to Deepgram's speech models, excluding Whisper models. The endpoint supports major APIs and requires minimal migration effort, enhancing compliance and performance for European applications.
Show More..

Deepgram has expanded its Nova-3 ASR model to support 10 new languages, including Greek, Romanian, and Malay, enhancing its multilingual capabilities. The update introduces Multilingual Keyterm Promp...ting, allowing for improved recognition of brand names and technical terms across languages without retraining. This advancement aims to improve accuracy and adaptability for global enterprises in diverse markets.
Show More..

Deepgram has launched its voice AI model integration with Amazon Connect and Amazon Lex, enhancing speech-to-text and text-to-speech capabilities. This integration offers ultra-low latency, high accu...racy, and scalability, improving customer service bots and compliance analysis. Early adopters report significant improvements in customer interactions. Deepgram will showcase this at AWS re:Invent.
Show More..

Deepgram has launched an integration of its STT, TTS, and Voice Agent APIs with Amazon SageMaker, enabling real-time speech processing directly through the SageMaker API. This integration addresses p...revious limitations in real-time speech applications on AWS, offering sub-second latency and seamless scalability. It enhances enterprise voice AI capabilities, particularly for contact centers and analytics teams, by simplifying infrastructure and improving responsiveness.
Show More..

Deepgram Launches Flux - The World's First Conversational Speech ...

Deepgram has achieved the AWS Generative AI Competency, highlighting its expertise in deploying generative AI technologies. This recognition underscores Deepgram's ability to deliver secure, complian...t, and scalable voice AI solutions, enhancing integration with AWS services like Amazon Bedrock and Amazon SageMaker. The competency assures customers of faster innovation and reduced costs within the AWS ecosystem.
Show More..

Deepgram plans to establish its APAC headquarters in Singapore following a strategic investment from EDBI. This move aims to enhance its regional presence, support customer and partner expansion, and... deepen its expertise in Asian languages and low-latency deployments. The investment aligns with Deepgram's strategy to meet growing demand for Voice AI in APAC.
Show More..

Deepgram, Fortanix, and Nvidia have partnered to introduce secure voice AI deployments for regulated industries. This collaboration leverages Nvidia's Confidential Computing-enabled GPUs and Fortanix...'s Confidential AI platform to keep audio data and model weights encrypted during real-time inference. The initiative targets sectors like healthcare, finance, and government, ensuring data security and compliance with strict regulatory requirements.
Show More..

Deepgram and Fortanix have partnered to offer a secure, on-premises voice AI solution using NVIDIA Confidential Computing. This collaboration ensures sensitive data and AI models remain protected dur...ing processing, meeting stringent regulatory requirements. The solution targets regulated industries, enabling private voice agents and enterprise-wide transcription with real-time performance.
Show More..

Dell Technologies, Penguin Solutions, and Deepgram have partnered to develop enterprise voice AI infrastructure. This collaboration focuses on Speech-to-Text, Text-to-Speech, and Voice Agent workload...s, utilizing Dell PowerEdge servers, PowerScale storage, and NVIDIA GPUs for real-time, low-latency performance. This partnership aims to enhance Dell's AI infrastructure offerings, targeting demanding enterprise AI deployments.
Show More..

Deepgram and IBM have partnered to integrate Deepgram's speech-to-text and text-to-speech capabilities into IBM's watsonx Orchestrate AI solution. This collaboration enhances IBM's offering with adva...nced voice recognition, enabling enterprises to automate operations and improve customer interactions. The integration supports diverse languages and accents, offering real-time transcription and natural-sounding speech, expanding Deepgram's reach through IBM's enterprise network.
Show More..

IBM and Deepgram have partnered to integrate Deepgram's speech-to-text and text-to-speech capabilities into IBM's watsonx Orchestrate AI solution. This collaboration enhances IBM's offerings with adv...anced voice recognition technology, enabling enterprises to automate operations and improve customer interactions. The integration supports multiple languages and dialects, offering real-time captioning and natural-sounding speech, which benefits sectors like healthcare and finance.
Show More..

IBM and Deepgram have formed a partnership to integrate Deepgram's speech-to-text and text-to-speech technology into IBM's watsonx Orchestrate platform. This collaboration aims to enhance enterprise ...AI workflows by embedding advanced voice capabilities, allowing for more natural interactions through spoken language. The integration supports multiple languages and dialects, offering improved accuracy and performance for use cases like automated customer service and voice-driven data entry.
Show More..

IBM and Deepgram have formed a partnership to integrate Deepgram's voice AI capabilities, including speech-to-text (STT) and text-to-speech (TTS), into IBM's watsonx Orchestrate platform. This collab...oration marks IBM's first voice AI partnership, enhancing enterprise AI interfaces by incorporating voice as a primary interaction layer. The integration aims to address complex audio challenges in industries like healthcare and finance, offering solutions for automated customer support and call analytics.
Show More..

Deepgram and IBM have partnered to integrate Deepgram's speech-to-text and text-to-speech capabilities into IBM's watsonx Orchestrate platform. This collaboration, marking Deepgram as IBM's first ded...icated voice partner, aims to enhance enterprise AI with real-time transcription, multilingual support, and natural-sounding speech synthesis. The integration supports various use cases, including automated customer support and voice-enabled data entry, expanding Deepgram's reach within IBM's enterprise ecosystem.
Show More..

IBM and Deepgram have partnered to integrate Deepgram's voice technology into IBM's watsonx Orchestrate platform. This collaboration will enhance the platform with real-time speech-to-text and text-t...o-speech capabilities, enabling enterprises to build voice-enabled agents and workflows. The integration aims to improve customer support, call analysis, and voice-driven data entry, addressing challenges like background noise and diverse accents.
Show More..

Deepgram and IBM have collaborated to enhance enterprise AI with advanced voice capabilities. This partnership aims to integrate Deepgram's voice recognition technology with IBM's AI solutions, offer...ing improved transcription and voice analytics for businesses.
Show More..

Deepgram has integrated its speech-to-text (STT) and text-to-speech (TTS) models with Amazon Connect and Amazon Lex. This integration enables real-time transcription, low-latency voice bots, and anal...ytics within AWS environments. The collaboration aims to enhance customer experience scenarios with higher accuracy and flexibility, allowing seamless deployment in AWS.
Show More..

Voximplant has integrated Deepgram's Voice Agent API into its cloud communications platform, enabling real-time Voice AI across various telephony environments. This partnership simplifies the deploym...ent of conversational voice agents by combining Voximplant's telephony infrastructure with Deepgram's advanced speech recognition and synthesis capabilities. The integration allows developers to focus on agent performance without managing complex telephony systems.
Show More..

Lindy has launched Gaia, an AI-powered voice agent platform, using Deepgram Flux to enhance phone conversations. Flux provides ultra-low latency and conversation awareness, allowing Gaia to manage ca...lls seamlessly. This integration transforms traditional calls into intelligent, automated interactions, supporting over 30 languages and enabling real-time actions. Flux's conversational speech recognition ensures fast, accurate, and natural dialogue.
Show More..

Deepgram's Nova-3 model received the 2025 Voice AI Technology Excellence Award from TMC for its exceptional accuracy, real-time multilingual transcription, and self-serve customization capabilities. ...The model excels in noisy environments and offers a leading word error rate, enhancing voice-powered applications across industries like healthcare and finance.
Show More..

Deepgram has been honored with the 2025 Customer Experience Innovation Award by CUSTOMER Magazine, recognizing its advancements in voice AI technology.