Voice AI in vogue: Nvidia releases new open transcription model
Nvidia released a free speech‑to‑text model that can transcribe recordings in up to 40 languages, letting you turn audio into text without a paid service.
Email Brief29 Oct 20266 stories
Hi, it's Poulami. Nvidia just released a free speech‑to‑text model that can transcribe recordings in up to 40 languages. I was surprised it works only on recorded audio, not live streams. Anyone who spends evenings typing out lecture notes will now get them in text automatically.
Nvidia released a free speech‑to‑text model that can transcribe recordings in up to 40 languages, letting you turn audio into text without a paid service.
Nvidia’s open Nemotron 3.5 Transcribe handles transcription, speaker recognition and translation in one model, so you can get multilingual subtitles and identify speakers without extra tools.
vLLM Semantic Router unveiled d3‑ultra, a 27 billion‑parameter multimodal decision model that reads text, images and video once and returns probability scores, enabling richer AI reasoning.
Google launched Gemini 4 Carbon, a frontier model that claims gains in coding, agentic tasks and cybersecurity, offering stronger performance for creators who need AI assistance.
Beijing announced a state‑owned data platform aimed at turning China’s massive data reserves into fuel for AI development, a move that could boost local AI services and content creation.
Research shows AI bots flooded a UK government consultation on disposable vapes, highlighting how automated responses can overwhelm policy processes and affect public discourse.
This is the email edition, sent every Thursday at 7 AM IST. Get it by email
Free, two ways. Pick one, or both.
Daily on WhatsApp
Five AI stories in plain language, posted to our WhatsApp Channel every morning. Free to follow, and only we can post there.
Weekly by email
Six of the most useful AI stories, every Thursday at 7 AM IST. Free. Unsubscribe any time.