Spaces:
Sleeping
Sleeping
| # Field Noise Samples | |
| Place `.wav` audio files here to enable realistic field-noise augmentation during training. | |
| ## Required Files (16kHz mono, any duration ≥5s) | |
| - `tractor_engine.wav` — diesel tractor idling or working | |
| - `wind_field.wav` — wind in open farmland | |
| - `livestock_ambient.wav` — cattle, goats, or chickens in background | |
| ## Suggested Sources | |
| - [Freesound.org](https://freesound.org) — search "tractor", "wind field", "livestock ambient" (filter by CC0 / CC-BY) | |
| - Field recordings from partner NGOs or agricultural organizations in Mali/Guinea | |
| ## Licensing Note | |
| Ensure all audio files are licensed for use in ML training datasets. | |
| CC0 (public domain) or CC-BY are preferred. | |
| ## Without Noise Files | |
| The augmenter will fall back to Gaussian noise only. | |
| Training will still work but model robustness to real-world conditions may be reduced. | |