RIPPLE: Generating Multi-Channel Phase, Not Recovering It

2026-07-30

AI Summary

This work addresses the critical issue in traditional multichannel audio and seismic waveform generation where independently recovering phase information per channel disrupts inherent inter-channel physical phase relationships, leading to spatial information loss that is often undetectable by magnitude-based metrics. To overcome this limitation, the study introduces an end-to-end phase generation framework that treats multichannel phase as a learnable variable for the first time. The approach initializes phase priors using the Griffin–Lim algorithm, integrates a rectified flow model, and incorporates an explicit multichannel phase loss function to enforce cross-channel phase consistency. Evaluated on Ambisonics environment transfer and seismic station transformation tasks, the method substantially improves phase coherence, reducing seismic S-wave polarization error from 57.3° to 33.8°—significantly outperforming conventional channel-independent phase recovery techniques.

Abstract

Generative models synthesize magnitude spectra with high fidelity, while phase is delegated to a recovery module---Griffin--Lim, a vocoder, or a latent decoder---applied independently to each channel. For multi-channel waveforms this delegation is costly: the physical content of spatial audio and three-component seismograms lives in the phase relationships between channels, precisely what channel-independent recovery cannot produce. The cost is also invisible, since the magnitude-based metrics common to both fields barely move when inter-channel phase coherence collapses---so a pipeline can discard the physical information in its output while still scoring well. We argue that phase should be generated, not recovered, and present RIPPLE (Rectified Inter-channel Phase with Prior-based LEarning), which reinterprets Griffin--Lim as a phase **prior** rather than a final estimator: initialized from the source phase, this prior carries the inter-channel structure to be preserved, and a rectified flow refines it toward the target under an explicit inter-channel phase loss. Tested on first-order ambisonics environment transfer and seismic cross-station translation---two physically unrelated domains---RIPPLE outperforms recovery-based pipelines on the coherence metrics that downstream analyses consume. The seismic case is decisive: across architecturally distinct generators, per-channel recovery leaves S-wave polarization error near the $57.3^\circ$ random expectation, whereas learned phase reduces it to $33.8^\circ$.

Problem

• multi-channel phase
• phase coherence
• spatial audio
• seismograms
• generative modeling

Innovation

• multi-channel phase generation
• inter-channel coherence
• phase prior
• rectified flow
• spatial audio and seismic signal processing