Audio Chunks for Building Floor Classification

Audio chunks for building floor classification, created by author kudukudu and last updated in September 2022. It includes subsets of 5-second and 10-second chunks derived from expert samples, as well as 10-second chunks from whole samples. The total number of rows and columns is unknown.

Use Cases

• Train an audio classifier to identify building floors using 5-second audio chunks from expert samples.
• Compare classification performance between models trained on 10-second chunks from expert samples versus whole samples.
• Analyze audio feature patterns across different chunk durations (5-second and 10-second) for floor-level identification.

Strengths

• Dataset provides multiple chunk durations (5-second and 10-second) for comparative analysis.
• Includes data derived from both expert samples and whole samples, offering varied sources.
• Last updated on 2022-09-08, providing a specific timestamp for the data's recency.

Limitations

• The dataset size, row count, and specific audio features (columns) are unknown, limiting pre-use assessment.
• Data is from 2022 and may not reflect current audio patterns or building environments.
• Lack of column details prevents understanding of the specific audio attributes available for modeling.

Provenance

Source

huggingface

Collection Method

Audio chunks obtained from expert samples and whole samples, with durations of 5 and 10 seconds.

Freshness

Last updated 2022-09-08.

Geography

Regionus (likely United States).