Omni-Embed-Audio: Leveraging Multimodal LLMs for Robust Audio-Text Retrieval
For practitioners needing robust audio retrieval beyond caption-style queries, OEA demonstrates that LLM backbones improve semantic understanding and hard negative discrimination.
Omni-Embed-Audio (OEA) uses multimodal LLMs for audio-text retrieval, achieving comparable text-to-audio performance to state-of-the-art M2D-CLAP while showing +22% relative improvement in text-to-text retrieval and +4.3%p HNSR@10 improvement in hard negative discrimination.
Audio-text retrieval systems based on Contrastive Language-Audio Pretraining (CLAP) achieve strong performance on traditional benchmarks; however, these benchmarks rely on caption-style queries that differ substantially from real-world search behavior, limiting their assessment of practical retrieval robustness. We present Omni-Embed-Audio (OEA), a retrieval-oriented encoder leveraging multimodal LLMs with native audio understanding. To systematically evaluate robustness beyond caption-style queries, we introduce User-Intent Queries (UIQs) - five formulations reflecting natural search behaviors: questions, commands, keyword tags, paraphrases, and exclusion-based negative queries. For negative queries, we develop a hard negative mining pipeline and propose discrimination metrics (HNSR, TFR) assessing models' ability to suppress acoustically similar distractors. Experiments on AudioCaps, Clotho, and MECAT show that OEA achieves comparable text-to-audio retrieval performance to state-of-the-art M2D-CLAP, while demonstrating clear advantages in two critical areas: (1) dominant text-to-text retrieval (+22% relative improvement), and (2) substantially superior hard negative discrimination (+4.3%p HNSR@10, +34.7% relative TFR@10), revealing that LLM backbones provide superior semantic understanding of complex queries.