Explore the signal, not just the sound. Inspect waveforms and frequency content, compare channels, mark important moments, and export a documented analysis—all on your device.
Load an audio file or a clearly labeled synthetic example to start an analysis.
◉Signal workspaceNo recording loaded
Explore recording
Waveform Drag to select an interval · click to seek
00:00.0000:00.00
Spectrogram Frequency on vertical axis · time on horizontal
Low frequencyHigh frequency at top
Time selection
Frequency spectrum Strongest FFT bin —
Peak level—
RMS level—
Near-clipped samples—
Detected silence regions—
Detected clip runs—
Selected duration—
Privacy first. Analysis stays in this tab. Source audio is not uploaded or stored by PixelOkay. Reports describe decoded audio and do not establish authenticity or chain of custody.
Inspect audio with useful, repeatable measurements
View the amplitude envelope and time–frequency content of an uploaded recording. Mark segments manually, flag low-energy regions and near-full-scale digital clipping, compare stereo channels and inspect a local FFT spectrum. Use exports to document observations without sending private recordings to a service.
Understanding the limits
The browser decodes compressed recordings and may resample them. Spectrograms use a Hann-window FFT; changing the window affects time/frequency resolution. Silence, clipping and abrupt energy-rise detection are heuristic indicators. They cannot prove tampering, identify a person or determine recording authenticity. For evidentiary work, preserve the original file and use validated professional procedures.
Using Audio Forensics Studio
Open supported audio or load the labeled synthetic example. Inspect channel-specific or combined spectral views, review candidate events and annotate intervals. Export descriptive reports or annotation JSON while retaining the original recording separately.
Analysis is descriptive, not authentication, speaker recognition, transcription or certified evidence handling. SHA-256 can match source files but does not establish chain of custody. Frequency levels are relative digital values rather than calibrated dB SPL. Decoding can resample; limits include 80 MiB files, 480 seconds, one or two channels and 40 million decoded samples. PDF uses browser printing.
Workflow and output choices for Audio Forensics Studio
Inspect supported audio with waveform and spectral views, review candidate events, add time-based annotations and export descriptive reports while retaining the original recording separately.
Annotation JSON preserves the notes, not a forensic chain of custody. A file fingerprint can help compare source files; PDF reporting uses browser printing. Keep the original recording unchanged when provenance matters.
Example workflow
Open a supported recording or the explicitly labeled synthetic example.
Review channel-specific or combined spectral views and select a time interval of interest.
Add an annotation, export a descriptive report or annotation JSON, and keep the original audio separately.
Authored example: On the labeled synthetic example, mark a short interval where a visible spectral change appears and write a neutral observation such as “energy increases.” This is not an authentication finding or fresh evidence report.
Limits and handling: Analysis is descriptive, not authentication, speaker recognition, transcription or certified evidence handling. Decoding may resample; frequency magnitudes are relative digital values, not calibrated dB SPL. File duration, size, channels and sample budgets apply.
Audio Forensics Studio questions
Does a matching SHA-256 fingerprint prove chain of custody?
No. It can show that two files have matching bytes; it does not establish who handled the recording or how it was collected.
Are frequency readings calibrated sound-pressure levels?
No. They are relative digital values, not calibrated dB SPL measurements.
