TR2026-098

The MERL Systems for DCASE 2026 Challenge Task 4

    •  Saijo, K., Masuyama, Y., Boeddeker, C., Wichern, G., Richter, J., Edo, T., Le Roux, J., "The MERL Systems for DCASE 2026 Challenge Task 4," Tech. Rep. TR2026-098, IEEE AASP Challenge on Detection and Classification of Acoustic Scenes and Events (DCASE Challenge), June 2026.
      BibTeX TR2026-098 PDF
      • @techreport{Saijo2026jun,
      • author = {{Saijo, Kohei and Masuyama, Yoshiki and Boeddeker, Christoph and Wichern, Gordon and Richter, Julius and Edo, Takahiro and Le Roux, Jonathan}},
      • title = {{The MERL Systems for DCASE 2026 Challenge Task 4}},
      • institution = {IEEE AASP Challenge on Detection and Classification of Acoustic Scenes and Events (DCASE Challenge)},
      • year = 2026,
      • month = jun,
      • url = {https://www.merl.com/publications/TR2026-098}
      • }
  • MERL Contacts:
  • Research Areas:

    Artificial Intelligence, Machine Learning, Speech & Audio

Abstract:

This technical report describes our spatial semantic segmentation of sound scenes (S5) systems for DCASE 2026 Challenge Task 4. Inspired by the top-ranked system in DCASE 2025 Task 4, we adopt a cascaded framework consisting of universal sound separation (USS) with source counting, source classification, and class-aware refine- ment. In the first stage, a TF-Locoformer-based USS model separates multi-channel mixtures into single-channel foreground and interference signals. Then, each separated signal is classified into one of 18 foreground classes or as interference. The separated fore- ground signals are further refined by another TF-Locoformer-based model conditioned on the predicted class labels and the observed mixture. Our best system achieves CA-PI-SDRi of 14.95 dB and mixture accuracy of 78.11% on the dev test set.