---
title: "Encoder-Side Neuron Identification and Amplification for Acoustic Perception in Large Audio-Language Models"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2607.11801v1"
permalink: "https://feed.astralor.com/items/itm_6f0bada0632454586edea81f"
published: "2026-07-13T16:53:08.000Z"
score: 81
channel: "ai"
tags: ["Audio-Flamingo-3","IAAN","Kimi-Audio","Qwen2.5-Omni","Inference-Time Intervention","Audio Encoder"]
---

# Encoder-Side Neuron Identification and Amplification for Acoustic Perception in Large Audio-Language Models

## AI commentary

> This paper is refreshing: instead of scaling compute or modifying architectures, it cleverly exploits activation differences of encoder neurons to achieve immediate acoustic perception gains. Especially valuable for LALM developers and audio researchers.

## At a glance

Large audio-language models (LALMs) often underperform on fine-grained non-semantic speech attributes such as speaker emotion. This paper introduces IAAN, a training-free and label-free method that scores each feed-forward neuron in the audio encoder by contrasting its activation on real waveform vs. a noise reference, then amplifies top-scoring neurons at inference. On ten non-semantic attributes, IAAN improves average accuracy by 25.7 points on Audio-Flamingo-3, 21.4 on Qwen2.5-Omni, and 9.7 on Kimi-Audio. Controlled experiments show both encoder locus and neuron-level selectivity are necessary; intervening after the encoder yields little gain. This work opens a new direction for inference-time methods that improve acoustic perception through neuron-level access to the encoder.

## Key points

- IAAN is the first method to perform neuron-level inference-time intervention inside the audio encoder without requiring retraining.
- It identifies key acoustic neurons by contrasting activations on real waveforms vs. a noise reference.
- IAAN significantly improves accuracy on non-semantic speech attributes across multiple LALMs, with up to 25.7 points gain.
- Experiments validate the necessity of encoder locus and neuron-level selectivity; post-encoder interventions yield little improvement.

## Full text

Large audio-language models (LALMs) often underperform on fine-grained, non-semantic attributes of speech, such as a speaker's emotion, despite strong performance on speech content. Improving this without the cost of retraining calls for an effective inference-time intervention, yet most existing methods intervene only after the audio encoder and operate at a relatively coarse granularity. The encoder itself, where acoustic information is first extracted from the waveform, remains largely unexplored, especially at the level of individual neurons. We introduce IAAN, Identifying and Amplifying Acoustic Neurons, a training-free and label-free method that scores each feed-forward neuron in the audio encoder by contrasting its activation on the real waveform with that on a noise reference lacking the real audio's acoustic information. IAAN then amplifies a small set of the highest-scoring neurons at inference. Across ten non-semantic speech attributes, IAAN improves average accuracy by 25.7 points on Audio-Flamingo-3, 21.4 on Qwen2.5-Omni, and 9.7 on Kimi-Audio. It also improves a model already explicitly fine-tuned to prioritize acoustic evidence. In controlled comparisons, both the encoder locus and neuron-level selectivity prove necessary for this gain. Intervening after the encoder, at the decoding side or inside the language model, yields little to no improvement, or even deteriorates accuracy. The improvement also depends on which specific neurons are amplified, not merely on their number, confirming that IAAN's acoustic score succeeds in identifying the neurons that matter. These results show that a small, precisely targeted intervention inside the audio encoder is an effective and largely untapped way to strengthen the acoustic understanding of LALMs, opening a new direction for inference-time methods that improve acoustic perception through neuron-level access to the encoder.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_6f0bada0632454586edea81f