Spatial Audio Event Detection And Localization Dataset

Featuring audio mixtures for sound event detection and localization (SELD) tasks, developed for the DCASE 2022 Challenge Task 3. It combines the STARSS22 dataset with synthetic mixtures, created by researchers from Tampere University and Sony R&D Center.

Use Cases

• Train models for sound event detection using spatial audio mixtures.
• Develop audio localization algorithms using multi-channel recordings.
• Benchmark SELD systems against the DCASE 2022 Task 3 challenge specifications.

Strengths

• Created for a major international audio research challenge (DCASE 2022).
• Combines real-world recordings (STARSS22) with synthetic mixtures.
• Developed by a collaboration between academic and industry research groups.

Limitations

• Specific audio formats, sample counts, and recording durations are not detailed in the provided input.
• The scope is specialized for the SELD task, limiting general audio applications.
• Annotations and metadata details are not specified in the provided description.

Provenance

Source

Tampere University and Sony R&D Center, via a copy from Zenodo.

Collection Method

Combination of real-world spatial audio recordings (STARSS22) and synthetically generated mixtures.

Freshness

2022-10-19