Detect Tampering and Edits in Audio Files
A free audio forensics tool for detecting signs of editing, splicing, or AI generation in recordings. Analyzes audio consistency and patterns to verify authenticity.
How to Use
1
Upload an audio file
2
Wait for analysis to complete
3
Review spectrogram for visual anomalies
4
Check noise floor and compression results
5
Read the verdict and confidence score
Detect audio tampering and deepfakes
Upload a file to start
MP3, WAV, OGG, FLAC
Upload a file to start
Analyzing...
Heuristic score: 0% (heuristic, not statistical)
How is this calculated?
This score combines 4 weighted heuristic checks. It is not a statistical confidence interval and does not replace professional forensic analysis.
score = (1 − raw / 6) × 100, where raw is the sum of weighted checks below:
- Noise-floor inconsistency (RMS coefficient of variation ≥ 0.80) — adds 0.5
- Multi-pass compression (low + high cutoff clusters) — adds 0.5
- ENF discontinuity (50/60 Hz hum drops or spikes) — adds 1.0
- Spectral cuts (flux > mean + 4σ) — adds 0.3 (any) up to ≈2.0 (many)
References: Grigoras (2007) for ENF analysis; standard spectral-flux onset detection.
Spectrogram
Noise Floor Analysis
Compression Artifacts
ENF Detection (50/60 Hz)
Spectrogram
Published
Updated
