Sonic Measurement Spaces: Quality Diversity Search and Sound Evolution

Björn Þór Jónsson (DataverseNO) published this dataset in 2025 to support research on Quality Diversity (QD) search algorithms in sound synthesis. It contains outputs from an Innovation Engine algorithm that evolves sounds using unsupervised dimensionality reduction and both full-reference and no-reference quality evaluation metrics.

Use Cases

• Benchmarking Quality Diversity (QD) search algorithms using the provided sonic measurement spaces
• Analyzing the effectiveness of unsupervised dimensionality reduction for structuring search containers
• Evaluating sound synthesis quality using the full-reference and no-reference evaluation approaches

Strengths

• Includes both full-reference and no-reference quality evaluation metrics
• Utilizes unsupervised dimensionality reduction for defining structured containers in QD search
• Generated using an Innovation Engine algorithm for autonomous sonic discovery

Limitations

• Search is confined to a single, structured search landscape, which may limit the breadth of evolution
• Lack of explicit column documentation or record counts in the provided metadata

Provenance

Source

Björn Þór Jónsson, DataverseNO

Collection Method

Synthetic/Algorithmic (Innovation Engine algorithm)

Freshness

Last updated October 2025.