PAL: Probing Audio Encoders via LLMs -- A Study of Information Transfer from Audio Encoders to LLMs
The integration of audio perception capabilities into Large Language Models (LLMs) has enabled significant advances in Audio-LLMs. Although application-focused developments, particularly in curating training data for specific capabilities e.g., audio reasoning, have progressed rapidly, the underlying mechanisms that govern efficient transfer of rich semantic representations from audio encoders to LLMs remain under-explored. We conceptualize effective audio-LLM interaction as the LLM's ability to proficiently probe the audio encoder representations to satisfy textual queries. This paper presents a systematic investigation on how architectural design choices can affect that. Beginning with a standard Pengi/LLaVA-style audio-LLM architecture, we propose and evaluate several modifications guided by hypotheses derived from mechanistic interpretability studies and LLM operational principles. Our experiments demonstrate that: (1) delaying audio integration until the LLM's initial layers establish textual context that enhances its ability to probe the audio representations for relevant information; (2) the LLM can proficiently probe audio representations exclusively through LLM layer's attention submodule, without requiring propagation to its Feed-Forward Network (FFN) submodule; (3) an efficiently integrated ensemble of diverse audio encoders provides richer, complementary representations, thereby broadening the LLM's capacity to probe a wider spectrum of audio information. All hypotheses are evaluated using an identical three-stage training curriculum on a dataset of 5.6 million audio-text pairs, ensuring controlled comparisons. Our final architecture, which incorporates all proposed modifications, achieves relative improvements from 10\% to 60\% over the baseline, validating our approach to optimizing cross-modal information transfer in audio-LLMs. Project page: https://ta012.github.io/PAL/
Code (0)
등록된 구현이 없습니다.
Methods 이 논문이 사용한 방법론
Similar Papers 제목 키워드 기반
USAD 2.0: Scaling Representation Distillation for Universal Audio Understanding
Audio encoders are critical to modern audio applications as large language models (LLMs) increasingly rely on a single encoder for diverse inputs. While self-supervised learning (SSL) has yielded strong domain-specific e…
Self-Supervised LearningUnmute the Patch Tokens: Rethinking Probing in Multi-Label Audio Classification
Although probing frozen models has become a standard evaluation paradigm, self-supervised learning in audio defaults to fine-tuning when pursuing state-of-the-art on AudioSet. A key reason is that global pooling creates …
Self-Supervised LearningAudio ClassificationSparse Autoencoders Make Audio Foundation Models more Explainable
Audio pretrained models are widely employed to solve various tasks in speech processing, sound event detection, or music information retrieval. However, the representations learned by these models are unclear, and their …
Self-Supervised LearningInformation RetrievalSound Event DetectionDo Audio-Visual Large Language Models Really See and Hear?
Audio-Visual Large Language Models (AVLLMs) are emerging as unified interfaces to multimodal perception. We present the first mechanistic interpretability study of AVLLMs, analyzing how audio and visual features evolve a…
Text GenerationImplications of Annotation Artifacts in Edge Probing Test Datasets
Edge probing tests are classification tasks that test for grammatical knowledge encoded in token representations coming from contextual encoders such as large language models (LLMs). Many LLM encoders have shown high per…
Memorization