Jiarui Hai

Experiments show that VoiceTrace achieves state-of-the-art performance on established semantic speech retrieval benchmarks, while substantially outperforming cascade-based approaches on VoiceTrace-Bench, demonstrating its effectiveness for both conventional semantic retrieval and the new hybrid retrieval setting.

Transcript-based shortcuts in dialogue models lead to a staggering drop in accuracy, revealing a critical flaw in current evaluation methods.

LLMs can now help you find that *specific* sound you're looking for, even if your training data is incomplete.