TR2025-045

Data Augmentation Using Neural Acoustic Fields With Retrieval-Augmented Pre-training

    •  Ick, Christopher, Wichern, Gordon, Masuyama, Yoshiki, Germain, François G, Le Roux, Jonathan, "Data Augmentation Using Neural Acoustic Fields With Retrieval-Augmented Pre-training", Tech. Rep. TR2025-045, Mitsubishi Electric Research Laboratories, Cambridge, MA, April 2025.
      BibTeX TR2025-045 PDF
      • @techreport{MERL_TR2025-045,
      • author = {Ick, Christopher; Wichern, Gordon; Masuyama, Yoshiki; Germain, François G; Le Roux, Jonathan},
      • title = {Data Augmentation Using Neural Acoustic Fields With Retrieval-Augmented Pre-training},
      • institution = {MERL - Mitsubishi Electric Research Laboratories},
      • address = {Cambridge, MA 02139},
      • number = {TR2025-045},
      • month = apr,
      • year = 2025,
      • url = {https://www.merl.com/publications/TR2025-045/}
      • }
  • MERL Contacts:
  • Research Areas:

    Artificial Intelligence, Machine Learning, Speech & Audio

Abstract:

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.