Universal-3 Pro by AssemblyAI

Universal-3 Pro is a next-generation, promptable speech language model. Unlike traditional automated speech recognition solutions, Universal-3 Pro takes in contextual prompts before processing, which improves the accuracy of its transcriptions and understanding of spoken language. It caters to specific content needs by recognizing and intelligently handling key aspects of speech like names, terminology, topics, and speech format. The tool surpasses conventional models by modifying output according to different contexts: providing verbatim transcriptions of clinical notes, tagging non-speech audio events, including disfluencies, recognizing informal speech and dialogue, and differentiating between speaker roles. It can also deal with code-switching, preserving the natural transition between languages such as English and Spanish. While being suitable for a number of applications, this tool shows potential for significant impact in areas like conversation intelligence, medical transcription, and contact centers, particularly where the capturing of nuanced speech components is critical.