The dataset is currently empty. Upload or create new data files. Then, you will be able to explore them in the Dataset Viewer.
MSR-UAV-Audio-16K2S
Dataset Summary
MSR-UAV-Audio-16K2S (Multi-Source Robust UAV Audio Dataset) is a standardized drone sound classification dataset composed of multi-source UAV audio recordings collected from public platforms (e.g., YouTube, Kaggle, Hugging Face) and private datasets.
All audio samples are:
- Mono channel
- 16 kHz sampling rate
- 2-second fixed-length clips
- Noise-augmented for robustness
The dataset is designed for robust UAV acoustic classification under diverse environmental conditions.
Dataset Characteristics
- Multi-source aggregated data
- Multiple UAV/drone categories
- Background noise augmentation
- Standardized preprocessing pipeline
- Fixed-length clips (2s)
- Mono 16kHz WAV format
Task
Audio Classification
Robust Drone Sound Recognition
Intended Use
This dataset is suitable for:
- UAV acoustic detection
- Drone classification research
- Surveillance systems
- Edge AI audio inference
- Robust environmental sound modeling
- Benchmarking noise-resistant classifiers
Data Processing
All raw audio was:
- Converted to mono
- Resampled to 16 kHz
- Segmented into 2-second clips
- Normalized
- Augmented with diverse environmental background noises
Noise sources include environmental and urban soundscapes to simulate real-world operating conditions.
Data Format
- Format: WAV
- Channels: Mono
- Sampling rate: 16,000 Hz
- Duration: 2 seconds per clip
Citation
If you use this dataset in your research, please cite:
MSR-UAV-Audio-16K2S: Multi-Source Robust UAV Audio Dataset, 16kHz, 2-second clips.
Limitations
- Data originates from heterogeneous sources.
- Background noise is synthetically augmented.
- May not fully represent all UAV acoustic signatures.
Future Work
- Real-world field recordings
- Expanded drone categories
- Multi-microphone recordings
- Longer temporal segments
- Downloads last month
- 55