AssemblyAI has released Universal-3.5 Pro, an advanced speech-to-text model designed to handle challenging audio conditions like noisy call centers and overlapping speakers. Unlike typical demos that use clean audio, Universal-3.5 Pro focuses on the difficult 20% of audio that impacts real-world accuracy. The new model offers improved entity accuracy for critical data like email addresses and phone numbers, and includes features like server-side Voice Focus for noise reduction and multichannel transcription to better manage overlapping speech. AI
IMPACT Enhances accuracy for AI-powered transcription services in noisy and complex audio environments.
RANK_REASON The item describes a new version of an existing speech-to-text model with improved features for specific use cases, rather than a frontier model release.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
