Top Updates đȘ
Krisp ships Voice Translation v3 with 96% accuracy in 61 languages and opens a self-serve developer API. (Krisp | Krisp)
Apple launches Siri AI at WWDC with multi-turn conversations and a standalone app powered by Gemini. (Apple)
Google launches Gemini 3.5 Live Translate for real-time speech translation across 70+ languages. (Google)
Mistral is raising ~âŹ3B at a âŹ20B valuation, nearly doubling since its September round. (TechCrunch)
Equal AI raises $30M Series B to scale Indiaâs voice-first AI assistant across a billion smartphones. (LiveMint)
NICE makes agentic AI the native architecture of its CX platform at NICE World 2026. (CMSWire)
Microsoft launches MAI-Voice-2, a TTS model supporting 10 languages and zero-shot voice cloning. (Blockchain Council)
AI voice scams surged 1,210% in 2025, needing just 3 seconds of audio to clone any voice. (Fox News)
Google will save search images and audio by default for AI model training. (The Verge)
AI ambient scribes cut physician burnout by 21 percentage points in a Mass General Brigham study. (Medical Daily)
MindBio delivers AI voice kiosks that detect intoxication and fatigue from speech patterns. (StreetWise Reports)
Top Gear asks whether AI voice control in cars is the next big thing or a waste of time. (Top Gear)
WSJ reports the job AI was supposed to kill now needs more humans than ever. (WSJ)
Voicegain hires a VP of Sales to push voice AI into healthcare call centers. (PRWeb)
Speechmatics named HackerNoonâs Company of the Week for speech AI innovation. (HackerNoon)
Voice AI adoption crosses an enterprise threshold in contact centers with measurable ROI. (CXToday)
India positions itself as the worldâs CX leader as voice AI reshapes its call center industry. (Express Computer)
Engineering Corner đ
Kyutai shows how RL post-training improves turn-taking and backchanneling in full-duplex voice models. (Kyutai)
Treble and Hugging Face launch FFASR, the first open benchmark for far-field speech recognition. (Newsfilecorp)
Red Hat publishes a guide to building a local voice agent with OpenShift AI. (Red Hat Developer)
DrivenData announces winners of âOn Top of Pasketti,â a childrenâs speech recognition challenge. (DrivenData)
Dev.to tutorial on extracting conversation intelligence from audio beyond simple dictation. (Dev.to)
Dev.to tutorial on building voice agents that send follow-up emails via Nylas. (Dev.to)
Blog tutorial covers building an ElevenLabs + n8n voice AI sales agent end to end. (whoisalfaz.me)
ParseJargon paper introduces real-time jargon translation for online meetings using LLMs. (arXiv)