By 
Introduction: A New Era for On-Device Voice AI
Deepgram has unveiled a groundbreaking initiative to deliver enterprise-grade speech recognition natively on PCs powered by Snapdragon® processors.
RELATED: IBM and Deepgram collaborate on enterprise Voice AI
By optimizing its Nova-3 speech-to-text model on the Qualcomm® Hexagon™ NPU within the Snapdragon X Series platform, the company is empowering developers and device manufacturers to deploy real-time voice experiences with unprecedented speed, enhanced privacy, and reliable performance—all without a cloud connection.
This marks a pivotal shift that paves the way for deeper voice integration across automotive, mobile, AI PCs, XR, industrial edge, IoT, and wearable devices.
The Shift: From Cloud Dependency to True Edge Computing
Traditional voice AI solutions have historically relied on a cloud-based architecture, where audio leaves the device, travels to the cloud for processing, and then returns with a response. This round-trip inevitably introduces latency and privacy vulnerabilities, restricting where voice AI can be effectively deployed. Deepgram is fundamentally disrupting this model by enabling speech recognition to occur entirely on-device. The result is a seamless, natural conversational experience—whether inside a moving vehicle, on an AI PC, within an XR headset, or at the network edge where connectivity is inconsistent or unavailable.
Unmatched Accuracy and Technical Superiority
Upendra Kulkarni, Vice President of Product Management at Qualcomm Technologies, emphasized the strategic value of the collaboration.
“We recognize that voice AI technologies serve different needs. That’s why we value our work with Deepgram – because accuracy, latency, reliability, and scalability matter for developers building automotive systems, healthcare applications, industrial solutions, and other mission-critical experiences. The on-device AI capabilities of Snapdragon processors combined with Deepgram’s enterprise-class voice AI uniquely bring those requirements together in a way that opens new possibilities for real-time voice experiences across the device ecosystem.”
Nova-3 represents a significant leap forward in enterprise-grade voice AI, delivering industry-leading accuracy even in challenging audio conditions. Key capabilities include:
- Real-time multilingual transcription: The first voice AI model to offer this feature natively.
- Instant self-serve customization: Enables immediate vocabulary adaptation without the need for costly and time-consuming model retraining.
- Superior accuracy metrics: Achieves a remarkable 6.89% word error rate on real-world production audio, representing a 24.7% lower error rate than the closest competitor.
Powering Mission-Critical Applications
Abe Pursell, Vice President of Business Development and Partnerships at Deepgram, reinforced the vision.
“The future of voice-first interaction is about meeting users wherever they are — in a vehicle, on a factory floor, inside an XR headset — without compromising on accuracy or responsiveness. To make that possible, you need enterprise-grade speech recognition that can deliver within the constraints of a device. We’re bringing Deepgram Nova-3 directly to the ecosystem of PCs with Snapdragon, providing a foundation to build voice into products that work wherever people happen to be, regardless of network conditions.”
Broad Ecosystem Impact and Future Outlook
This strategic integration removes the traditional barriers of connectivity and privacy, unlocking a new generation of intelligent applications.
By offering high-performance, low-latency voice AI at the edge, Deepgram and Qualcomm are collectively accelerating the adoption of voice-first interfaces across diverse sectors—from industrial automation and healthcare to next-generation AI PCs and immersive extended reality.
