ENF · mains hum · audio
What does the mains hum in this recording reveal?
Anything recorded near mains-powered equipment picks up a faint hum at the power-grid frequency, 50 Hz across most of the world and 60 Hz across the Americas. That hum wobbles very slightly over time, and the wobble is the same everywhere on a grid at a given moment. This reads the hum out of your file: it tells you the nominal frequency (which points to the grid region), shows how the trace drifts, and flags sudden jumps that can mark where a recording was cut. Read it as a lead, not proof.
Runs entirely in your browser. Your file is never uploaded or stored.
Decoding and tracking the mains hum…
A genuine, unbroken recording gives a smooth, continuous trace that wanders inside a narrow band around the nominal frequency (a real grid stays within about 0.6 Hz). A sharp step in the line, or a stretch where it scatters, is where the hum's phase breaks, which is what a cut or an inserted segment leaves behind.
What ENF can and cannot tell you
| Question | What ENF gives you |
|---|---|
| Which grid region? | Coarse. The nominal (50 vs 60 Hz) narrows it to a set of countries; distinguishing grids within that set is research-grade and not done here |
| Was it cut or spliced? | Yes, when the hum is present: a phase discontinuity in the trace marks an edit point |
| Exactly when was it recorded? | Not in a browser. Dating means matching this trace against a database of the grid's recorded frequency over time, which this tool does not have |
| Is there no hum at all? | Then there is nothing to read: the recording was made away from mains, heavily filtered, or the hum was removed |
How it works
The power grid runs at a set frequency, but supply and demand never balance perfectly, so the real frequency drifts by small fractions of a hertz from second to second. Because the grid is one interconnected machine, that drift is effectively identical everywhere on it at the same instant. Grigoras named this the electric-network-frequency criterion for forensic audio. The tool downsamples your recording, isolates the band around the nominal, and estimates the exact hum frequency in each short window to draw the trace. Where the recording is continuous the trace is smooth; where it was edited the hum's phase jumps, which Nicolalde-Rodriguez and colleagues showed can be read as an authenticity check. Hajj-Ahmad, Garg and Wu showed the statistics of the trace also carry a signature of the grid region it was recorded in.
What fools it
The whole method depends on a hum being captured in the first place. A recording made outdoors, on a battery device far from wiring, or through a microphone with a strong low-cut filter may carry no usable ENF, and this tool will say so rather than invent a result. Aggressive noise reduction can also strip the hum, and re-encoding at a low bitrate can bury it. A single clean trace is a strong sign of continuity, but a broken one should be corroborated with the spectrogram, the silence and gap finder and the waveform view before you conclude anything.
Sources
- Grigoras, C. (2005). Digital Audio Recording Analysis: The Electric Network Frequency Criterion. International Journal of Speech, Language and the Law / IAFPA.
- Nicolalde-Rodriguez, D. P., Apolinario, J. A. & Biscainho, L. W. P. (2010). Audio Authenticity: Detecting ENF Discontinuity With High Precision Phase Analysis. IEEE Transactions on Information Forensics and Security 5(3):534-543. DOI: 10.1109/TIFS.2010.2051270
- Hajj-Ahmad, A., Garg, R. & Wu, M. (2015). ENF-Based Region-of-Recording Identification for Media Signals. IEEE Transactions on Information Forensics and Security 10(6):1125-1136. DOI: 10.1109/TIFS.2015.2398367