OpenMOSS Releases MOSS-Audio: An Open-Source Foundation Model for Speech, Sound, Music, and Time-Aware Audio Reasoning

OpenMOSS, MOSI.AI, and Shanghai Innovation Institute released MOSS-Audio, an open-source audio foundation model.

The model comes in four variants and can handle transcription, speaker and emotion analysis, sound and music understanding, and audio QA.

MOSS-Audio also supports time-grounded questions and multi-hop reasoning over audio, aiming to unify several specialized tools into one system.