Onset detector

Press Start button at the bottom to begin playing the audio file. The white-on-black plot below the grid shows how the frequency content of this excerpt from a Mozart piano piece is changing over time (low frequencies toward the left of the x-axis and high frequencies to the right, and amplitude or strength on the y-axis).

Changes in the frequency content are used to estimate possible locations of onsets. When the audio excerpt reaches the end, all estimated onsets are plotted in the grid. The x-axis of the grid represents approximately 30 sec in time; the y-axis is the strength (loudness) of the detected onset at a particular time. For example, you might have heard two loud chords toward the excerpt's end – these are represented by the two 4-block orange bars right of center in the grid.

This project was made possible with the help of Tone.js and NexusIU. The demo was coded by Casey Ching as part of Dr. Collins' PSYC397 Seminar in Music Computing and Psychology.