TAU Urban Acoustic Scenes 2019 Openset, Evaluation dataset

Tampere University provides a 20-hour audio dataset for evaluating acoustic scene classification models. It consists of 10-second segments recorded in 10 distinct urban acoustic scenes, such as airports, metro stations, and parks. The dataset also includes material for an 'unknown' scene category.

Use Cases

• Benchmarking acoustic scene classification models based on the 10 defined urban scene categories.
• Developing open-set audio recognition systems using the provided 'unknown' scene material.
• Training urban sound event detection models on labeled 10-second audio segments.

Strengths

• Contains 20 hours of total audio data for model training and evaluation.
• Provides a defined set of 10 distinct urban acoustic scene categories for structured classification tasks.
• Includes material for an 'unknown' scene, supporting open-set evaluation scenarios.

Limitations

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

Provenance

Source

Tampere University