Data Augmentation Using Neural Acoustic Fields With Retrieval-Augmented Pre-training
This work addresses data augmentation for acoustic simulation in specific ICASSP workshop tasks, appearing incremental as it builds on existing neural acoustic field methods with retrieval-augmented pre-training.
The authors tackled room impulse response (RIR) estimation for data augmentation by pre-training a neural acoustic field on external geometry-RIR pairs and adapting it to target rooms using enrollment data, achieving unspecified results for RIR prediction and speaker distance estimation tasks.
This report details MERL's system for room impulse response (RIR) estimation submitted to the Generative Data Augmentation Workshop at ICASSP 2025 for Augmenting RIR Data (Task 1) and Improving Speaker Distance Estimation (Task 2). We first pre-train a neural acoustic field conditioned by room geometry on an external large-scale dataset in which pairs of RIRs and the geometries are provided. The neural acoustic field is then adapted to each target room by using the enrollment data, where we leverage either the provided room geometries or geometries retrieved from the external dataset, depending on availability. Lastly, we predict the RIRs for each pair of source and receiver locations specified by Task 1, and use these RIRs to train the speaker distance estimation model in Task 2.