Nature Machine Intelligence published a piece about the development of general auditory intelligence for machines to listen and speak. The description states that computer listening encompasses understanding, processing, and generating speech, environmental sounds, and music.
According to the publisher, the direction goes beyond traditional approaches and aims to leverage the capabilities of fundamental models for more complete understanding, natural generation, and interaction that is more human-like.
Why this matters: audio contains semantic, emotional, and contextual signals, so it is considered an important part of natural and embodied machine intelligence. The full text of the publication and details about specific systems are not provided in the supplied package.
editorial commentary
Why it matters
Probable consequence: audio will be more frequently regarded as an independent channel for systems that require semantic, emotional, and contextual understanding. The next observable signals will be specific methods, comparative evaluations, and demonstrations of such systems in the full text of the publication or in subsequent research. Substantial uncertainty arises from the fact that current information is presented only as metadata and does not allow judgment of practical readiness of approaches.