Video scrubber

A video recording holds two streams of data: a sequence of pictures and a sound track. This app lays both out along the same time axis, so that you can see where things happen without watching and listening through the whole recording. The videogram squeezes each video frame into a single column of pixels, the motiongram does the same with the changes between frames, and the spectrogram shows how the frequencies in the sound change over time. Click any of them to jump to that moment in the video.

Privacy: files you choose and clips you record stay in your browser. They are analysed on your own device and are never uploaded or stored anywhere.

Recording to analyse
No file chosen

Most browsers read MP4 files with H.264 video and WebM files. Files up to 300 MB and 10 minutes long are accepted; longer files take longer to analyse.

The browser asks for permission to use the camera and the microphone. The camera is switched off as soon as the clip is finished, and the clip is then analysed like a file. Save the clip only if you want to keep a copy on your own device.

Display
Picture
Frequency axis

Choose a video file or record a clip to begin.

Videogram

Sound

The waveform: how far the sound pressure swings at each moment. Loud moments are tall, silence is a flat line.

The spectrogram: time runs to the right, frequency upwards, and colour shows how strong each frequency is, from dark (80 dB below the loudest moment) to yellow (the loudest). A clap is a short vertical stripe across many frequencies; a sung or spoken vowel is a stack of horizontal lines, its harmonics. The logarithmic axis gives each octave the same height, closer to how we hear pitch; the linear axis gives each hertz the same height.

All panels share one time axis, in seconds. Click a panel to jump to that moment; the red line follows the playback. Drag across a panel to mark a stretch of time, and playback then loops within it until you clear the loop.

Things to try: record yourself clapping three times, then waving a hand, then clapping again, and find the claps in both the motiongram and the spectrogram. Does the hand movement before each clap show up in the motiongram? Load a music video and look for the cuts in the videogram and the beat in the waveform. Record yourself speaking a sentence and then singing it, and compare the spectrograms: where are the harmonics clear and steady, and where do they glide? Switch between the logarithmic and the linear frequency axis and see where the voice ends up.

Based on VideoScrub and the Musical Gestures Toolbox.