MMAR: AChallenging Benchmark for Deep Reasoning in Speech, Audio, Music, and Their Mix

Open in new window