Sound Detection Accuracy (SDA) Disclaimer

Last Updated: February 9, 2026

Microphone-Related Factors

Physical Condition and Placement

The quality of sound detection is directly dependent on your device's microphone condition and placement:

• Dirty or Blocked Microphone: Dust, lint, or debris in the microphone opening can significantly reduce detection accuracy
• Physical Obstructions: Phone cases, screen protectors, or your hand covering the microphone can muffle or block sound input
• Microphone Damage: Worn or damaged microphone hardware may produce distorted audio that affects detection quality
• Optimal Positioning: For best results, ensure the microphone has clear, unobstructed access to ambient sounds

Recommendation: Regularly clean your device's microphone opening and ensure it remains unobstructed during use.

Device-Specific Variations (Android)

Microphone Hardware Differences

Android devices exhibit significant variation in audio capture characteristics:

• Non-Standardized dBFS Levels: Different Android devices produce different decibel full scale (dBFS) readings for the same real-world sound due to:
  • Varied microphone hardware components across manufacturers
  • Different built-in gain settings and amplification levels
  • Manufacturer-specific audio processing algorithms
  • Device-specific noise cancellation and automatic gain control (AGC) systems
• Threshold Calibration Challenges: Because audio level measurements are not consistent across devices, universal threshold settings cannot be applied. Detection sensitivity may vary between different Android phone models, even when using the same version of FLOOIE.
• Detection Consistency: While FLOOIE attempts to adapt to different devices, some Android phones may experience higher false positive or false negative rates compared to others.

Note: iOS devices generally provide more consistent audio input characteristics due to standardized hardware, but variations can still occur.

Machine Learning Model Limitations

Training and Background Noise

FLOOIE's detection system is powered by a TensorFlow Lite machine learning model:

• Noise Handling: The model has been trained on datasets that include background noise mixed with coughs and sneezes to improve its ability to distinguish target sounds from environmental noise
• Not 100% Accurate: Despite extensive training, the system cannot achieve perfect accuracy. It is designed to suppress background noise and detect relevant sounds, but this process is not infallible
• Environmental Challenges: Particularly noisy environments (construction sites, loud music, crowds, traffic) may reduce detection accuracy
• Similar Sounds: Sounds that acoustically resemble coughs or sneezes (clearing throat, hiccups, certain speech patterns) may occasionally trigger false detections

Common Sound Detection Limitations

General Audio Detection Challenges

Like all audio-based detection systems, FLOOIE is subject to inherent limitations:

• Ambient Noise Interference: High levels of background noise can mask or distort target sounds
• Distance Limitations: Detection accuracy decreases with distance from the sound source
• Sound Overlap: Multiple simultaneous sounds can interfere with accurate detection
• Acoustic Environment: Room acoustics, echoes, and reflections can affect sound characteristics
• Volume Variations: Very quiet coughs or sneezes may not be detected; very loud background sounds may cause false positives
• Audio Processing Latency: There may be a slight delay between when a sound occurs and when it is detected and reported

Detection Performance Expectations

What You Should Know

• False Positives: FLOOIE may occasionally detect sounds that are not coughs or sneezes
• False Negatives: FLOOIE may occasionally fail to detect actual coughs or sneezes, particularly if they are quiet, distant, or masked by other sounds
• Proximity Estimation: Distance calculations are approximate and may not be precise in all situations
• Ongoing Improvement: We continuously work to improve detection accuracy through model updates and refinements

User Responsibilities

To optimize FLOOIE's performance:

1. Maintain Your Microphone: Keep the microphone clean and unobstructed
2. Understand Device Limitations: Be aware that your specific device model may perform differently than others
3. Use Appropriate Expectations: Treat FLOOIE as an awareness tool, not an infallible detection system
4. Provide Feedback: Report persistent detection issues to help us improve the system
5. Combine with Other Precautions: Do not rely solely on FLOOIE for health and safety decisions

Limitation of Liability

By using FLOOIE, you acknowledge and accept that:

• Detection accuracy may vary based on the factors described in this disclaimer
• We cannot guarantee specific performance levels across all devices and environments
• You use FLOOIE at your own risk and with full understanding of its limitations
• We are not liable for any consequences resulting from inaccurate detections, missed detections, or reliance on the app

Updates may improve performance, but the fundamental limitations described in this disclaimer will continue to apply to some degree.

Contact Us

If you experience persistent detection issues or have questions about accuracy:

Email: support@flooie.app
Website: www.flooie.app