According to ElevenLabs (@elevenlabsio), the company has launched Scribe v2 Realtime, an ultra-low latency Speech to Text model specifically optimized for agentic use cases. The new AI model addresses common challenges in speech recognition, including poor audio quality, diverse accents, and the accurate transcription of identifiers such as IDs and emails. This release highlights a major advance in real-time AI transcription technologies, offering significant opportunities for businesses in customer service automation, contact centers, and voice-driven enterprise applications. The improved accuracy and speed of Scribe v2 Realtime can streamline workflows, reduce operational costs, and enhance user experience in scenarios that demand instant and reliable speech recognition (Source: ElevenLabs Twitter, Nov 13, 2025).

According to Andrew Ng (@AndrewYNg) on Twitter, a new course titled 'Building Live Voice Agents with Google’s Agent Development Kit (ADK)' offers practical instruction for developers to build advanced voice-activated AI assistants. The course, taught by Google’s @lavinigam and @sitalakshmi_s, demonstrates how ADK's modular components and web-based debugging tools can streamline the creation of AI agents capable of real-world tasks like gathering AI news, scripting podcasts, and managing multi-speaker audio production. The curriculum emphasizes constructing agents that maintain context, implement guardrails, and efficiently handle audio streaming with low latency. This presents a significant business opportunity for organizations aiming to launch real-time, reliable AI voice applications in industries such as media, customer service, and content creation (Source: Andrew Ng, Twitter, Oct 15, 2025).