- The AI Voice Newsletter
- Posts
- Deepgram Teams with Fortanix, NVIDIA for Secure Voice AI
Deepgram Teams with Fortanix, NVIDIA for Secure Voice AI
🔊 Soundcheck
Secure on‑prem voice AI for regulated industries
Astreya Unifies Voice AI and ITSM
Twilio Seizes AI Voice Moment
Alibaba’s voice AI tops global speech chart
Read time: 4 minutes
🔥 Hot Mic
Big moves, deep dives, and standout stories.
Deepgram partners with Fortanix and NVIDIA to deliver on‑premises, confidential voice AI for highly regulated sectors. Deepgram has teamed up with Fortanix and NVIDIA to bring private, on‑premises voice AI to industries with strict security requirements. Their new solution combines Deepgram’s enterprise‑grade voice models with Fortanix Confidential AI and NVIDIA Confidential Computing to keep both audio data and proprietary model weights encrypted during active inference. This hardware‑isolated setup ensures that sensitive information never touches exposed infrastructure, enabling real‑time voice AI use in environments like healthcare, finance, and government. The deployment empowers organisations to build voice assistants, transcription layers, and internal voice tools without sacrificing data sovereignty or performance.
Key Points:
On‑prem voice AI uses hardware‑isolated, encrypted environments.
Audio and model weights remain encrypted even during active inference.
Powered by Deepgram’s voice models, Fortanix Confidential AI, NVIDIA Confidential Computing.
Supports regulated industries: healthcare, finance, government.
Takeaway: Deepgram’s partnership with Fortanix and NVIDIA marks a milestone: voice AI can now be deployed securely on‑premises with confidential computing, giving regulated enterprises the real‑time performance they need without compromising data or model privacy.
Astreya integrates 3CLogic’s voice AI into ServiceNow, creating a seamless, AI-powered IT service desk. Astreya has expanded its AI-first service desk by integrating 3CLogic’s voice AI and contact center tools directly into ServiceNow IT Service Management. This moves voice from a fragmented channel to a core, intelligent part of the support interface. Real-time transcription, guided agent actions, generative summaries, and quality monitoring arrive baked into the workflow from day one. Agents now work in a single environment where voice interactions turn into searchable, context-rich assets, boosting efficiency and user experience.
Key Points:
Voice AI and contact center functions now embedded into ServiceNow ITSM.
Live transcription, next-best‑action guidance, and generative summaries included.
AI-driven quality monitoring and coaching operate at scale.
Unified agent workspace reduces tool switching and speeds resolution.
Takeaway: By embedding voice AI directly into ServiceNow, Astreya transforms service desks into unified, intelligent environments—making every call a smarter, more actionable asset.
Twilio is ramping up its Voice AI offering and showing impressive market strength, making it clear that voice-driven AI is quickly becoming a key battleground for SaaS platforms.
Analysts note Twilio’s new Voice AI tools—like Conversation Relay—are helping the company move up the value chain, with meaningful revenue gains expected from AI-powered voice interactions.
This push into Voice AI coincides with strong stock performance, with analysts lifting price targets amid growing optimism that Twilio stands poised to capture structural growth from the AI revolution. Continued innovation and customer adoption suggest this is just the beginning.
Key Points:
Twilio’s Voice AI suite—including Conversation Relay—is rapidly expanding.
Analysts expect Voice AI to deliver higher per-call revenue and margins.
Voice is poised to become a larger slice of Twilio’s total revenue mix.
Market sees Twilio as foundational provider of AI-driven voice infrastructure.
Takeaway: Twilio’s strategic shift into Voice AI is resonating with both customers and analysts, setting it up to capitalize significantly on demand for conversational AI—a growth opportunity that could reshape its positioning in cloud communications.
Alibaba’s Fun‑Realtime‑TTS‑Preview top‑five ranking reflects its strength in multilingual, accent‑aware speech AI. Alibaba’s Tongyi Lab has unveiled a powerful new voice AI model, Fun‑Realtime‑TTS‑Preview, which ranked fifth overall on the Artificial Analysis Speech Arena global leaderboard. It was the only Chinese‑engineered system to make the top five, outperforming offerings from OpenAI and xAI. The model supports over 30 languages, seven major Chinese dialects, and more than 20 regional accents, demonstrating notable strength in handling diverse speech patterns. Alongside its text‑to‑speech innovation, Alibaba’s Fun‑Realtime‑ASR model achieved the lowest word error rate on the same benchmark, counting just 1.8% errors. Positioned for enterprises, the system includes customization tools for healthcare and finance, like converting spoken medical notes into structured records, signaling both technical prowess and practical relevance.
Key Points:
Fun‑Realtime‑TTS‑Preview ranked fifth globally in Speech Arena benchmark
Only Chinese‑built voice system in the global top five
Supports 30+ languages, seven Chinese dialects, and 20+ regional accents
Fun‑Realtime‑ASR achieved 1.8% word error rate, topping the recognition chart
Takeaway: Alibaba’s latest voice AI deliver a compelling proof point: exceptional performance in multilingual and regional-dialect speech, combined with enterprise-ready tools, positions it as a serious global contender.
🎙️ Mic Drop
What else is making noise in voice AI.
Infobip, with Omdia, offers strategic enterprise whitepaper on deploying and future-proofing Voice AI for customer experience and ROI. (finance.yahoo.com)
The FBI alerts public to a surge in AI-driven voice cloning scams, underscoring urgent security and trust concerns for voice AI. (click2houston.com)
Consumers recount real experiences with AI voice-cloning scams; public awareness rises, increasing pressure for technical safeguards. (wsls.com)
Study reflects $2.3B in elderly losses to scams using AI-cloned voices, showing business-critical need for authentication solutions. (yahoo.com)
Voice assistant market forecasted to expand to $32.5B by 2035, presenting opportunities for developers building smart home integrations. (sphericalinsights.com)
MSI integrates an advanced AI voice assistant into gaming PCs, illustrating new use cases in consumer technology and entertainment. (hothardware.com)
Greece adopts voice AI for multilingual digital tourism, expanding accessibility and offering a model for global travel and hospitality. (travelandtourworld.com)
Jotform enables users to build and configure surveys using conversational AI, streamlining workflows and voice-based data entry. (aimagazine.com)
Clayfin acquires Louie Voice; new banking integrations emphasize end-to-end transactions by natural speech, boosting accessibility. (thewire.in)
Kaltura recognized as an 'Exemplary Provider' for conversational AI, reflecting innovation in adaptive voice-driven digital avatars. (stocktitan.net)