Universal-3 Pro by AssemblyAI

Universal-3 Pro by AssemblyAI is a first of its kind promptable speech language model. This model is designed to deliver transcriptions that understand and reflect the specific content being communicated. It offers features that enhance the quality and value of transcriptions, enabling better interpretation and representation of spoken information.

Universal-3 Pro processes content by taking context about names, terminology, topics, and formats before processing. It uses this context to generate transcriptions that closely mirror the specific content of the conversation or speech being transcribed, ensuring that important details and nuances are captured accurately.

Yes, Universal-3 Pro can identify speaker roles. It offers role-specific prompts to guide the transcription process, and each speaker's role is correctly captured and labelled in the resulting transcription. This feature ensures that the dialogue structure is maintained and the roles of the speakers are clear in the transcript.

Universal-3 Pro handles speech disfluencies by accurately transcribing content that includes elements like stutters and informal speech. This capability ensures that the transcription retains the natural rhythm, flow, and other significant data from the conversation. This includes capturing fillers, repetitions, restarts, stutters and informal speech, preserving them accurately in the transcript.

Yes, Universal-3 Pro is capable of transcribing bilingual conversations. It is adept at handling language code-switching and preserves natural transitions between languages in bilingual conversations. This means it can smoothly switch between different languages while transcribing, mirroring the natural flow of bilingual dialogue.