Meta beats OpenAI and Google at real-time transcription
What this changes for captions, meetings and voice support.
Meta introduced Muse Voice Transcribe, a system that turns speech into text while someone is still speaking. In a real-time transcription benchmark, it ranked ahead of systems from OpenAI and Google. It can also track more than 20 speakers and uses reinforcement learning to decide how long to listen before recording each word.
This could be useful for creating captions during a conversation, recording meetings, or supporting voice-based customer service. Tracking many speakers may help when several people are taking part. Still, the reported result comes from a benchmark; the source does not explain how the system performs in each everyday use.
To get started, watch for whether Muse Voice Transcribe becomes available in captioning, meeting, or customer service tools, and how it works in those services. The source does not say how to access it or report practical results in these settings. For now, the concrete information is its benchmark performance and its ability to track more than 20 speakers.
The link opens the original content; some require a subscription.
Go further on INEMA.CLUB
More news
- 15/0912 new AIs expected in October 2026What is rumor and what is already a verified release, in one guide.
- 02/09Humanoid robots getting more realistic: the UBTECH caseMore companies are showing human-like robots, and one of them fools you at first glance.
- 02/10Raio-X de Margem helps clinics and salons track losses and goalsThe dashboard compares results before and after, and tracks losses, targets, and attributable bonuses in a pilot with 1,302 clinics and salons in Batel.
- 02/10NVIDIA launches an open platform to control AI agentsThe platform lets users monitor, limit, and stop AI agents that exceed their defined permissions.
- 02/10FEP — Prompt Engineering now in English and SpanishThe FEP — Prompt Engineering course is now available in English and Spanish.