TAU-NIGENS Spatial Sound Events 2021: Synthetic Scenes with Directional Interference

Tampere University created the TAU-NIGENS Spatial Sound Events 2021 dataset for the DCASE 2021 Challenge. It contains multiple spatial sound-scene recordings with sound events spatialized using real room impulse responses from various rooms. The dataset introduces more challenging conditions than previous iterations, including a maximum polyphony of 3 target events and the inclusion of directional interference events.

Use Cases

• Training sound event localization models based on direction-of-arrival and trajectory labels.
• Developing models for polyphonic sound scene analysis based on scenes with up to 3 simultaneous target events.
• Benchmarking audio source separation algorithms against directional interference events sourced from 'engine', 'fire', and 'general' sound classes.
• Evaluating spatial audio processing techniques using the provided microphone array and first-order Ambisonics recording formats.

Strengths

• Uses real spatial room impulse responses captured in multiple rooms with various shapes, sizes, and acoustical properties.
• Includes both static and moving sound sources, with time-variant RIRs used for moving sources.
• Represents the third iteration of a series, introducing more natural temporal distributions and directional interference events for increased realism.

Limitations

• Row count and dataset size are unknown, which may limit suitability assessment for large-scale training.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.

Provenance

Source

Tampere University, created by Archontis Politis.

Collection Method

Synthesized from the NIGENS general sound events database, spatialized using real captured room impulse responses.