MusicRFM discovers interpretable concept directions in music model hidden states using RFM probes and injects them at inference to steer generation toward desired musical properties without retraining.
Haven Kim, Zachary Novack, Weihan Xu, Julian McAuley, and Hao-Wen Dong
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
abstract
Understanding how large audio models represent music, and using that understanding to steer generation, is both challenging and underexplored. Inspired by mechanistic interpretability in language models, where direction vectors in transformer residual streams are key to model analysis and control, we investigate similar techniques in the audio domain. This paper presents the first study of latent direction vectors in large audio models and their use for continuous control of musical attributes in text-to-music generation. Focusing on binary concepts like tempo (fast vs. slow) and timbre (bright vs. dark), we compute steering vectors using the difference-in-means method on curated prompt sets. These vectors, scaled by a coefficient and injected into intermediate activations, allow fine-grained modulation of specific musical traits while preserving overall audio quality. We analyze the effect of steering strength, compare injection strategies, and identify layers with the greatest influence. Our findings highlight the promise of direction-based steering as a more mechanistic and interpretable approach to controllable music generation.
representative citing papers
A native quantized runtime runs Stable Audio 3 on commodity and Pi hardware with 8-bit quality within seed noise, 7× faster cold start, and bounded in-graph taste steering.
Hallucination information is linearly separable in Whisper activations and SAE latents; SAE steering reduces hallucination rates from 72.63% to 14.11% (small) and 86.88% to 27.33% (large-v3) on non-speech audio with small WER impact.
Activation steering with Gram-Schmidt orthogonalization enables disentangled, deterministic control of pitch and duration attributes in the Multitrack Music Transformer without retraining.
citing papers explorer
-
Steering Autoregressive Music Generation with Recursive Feature Machines
MusicRFM discovers interpretable concept directions in music model hidden states using RFM probes and injects them at inference to steer generation toward desired musical properties without retraining.
-
A Quantized Native Runtime for On-Device Semantic Audio Generation
A native quantized runtime runs Stable Audio 3 on commodity and Pi hardware with 8-bit quality within seed noise, 7× faster cold start, and bounded in-graph taste steering.
-
Whisper Hallucination Detection and Mitigation via Hidden Representation Steering and Sparse AutoEncoders
Hallucination information is linearly separable in Whisper activations and SAE latents; SAE steering reduces hallucination rates from 72.63% to 14.11% (small) and 86.88% to 27.33% (large-v3) on non-speech audio with small WER impact.
-
Latent Space Disentanglement via Activation Steering for Interpretable Attribute Control in Symbolic Music Generation
Activation steering with Gram-Schmidt orthogonalization enables disentangled, deterministic control of pitch and duration attributes in the Multitrack Music Transformer without retraining.