Loudspeaker time alignment test on headphones

This experiment allows you to test how time alignment between bass and treble affects the sound on headphones. In real loudspeakers, the tweeter and woofer could be at different distances from the listener and cause time differences. However, this test accentuates audible effects due to digital summing, acoustic summing is much more complex with different directivity of drivers and effects of the room.

The signal is split into high-passed (treble) and low-passed (bass) bands using digital filters. You can then add a time delay to the high-pass band to simulate moving the tweeter closer (negative delay perception) or farther away (positive delay). Signal is summed back in digital domain after the delay block.

Available crossover filters include 2nd and 4th order Linkwitz-Riley filters as well as 3rd order Butterworth filters. The delay can be adjusted from 0 to 2 milliseconds offset distances.

One millisecond measn about 1ft, or 34.3 cm distance at the speed of sound (343 m/s). 1cm is ~0.03 milliseconds, 1" is ~0.07 milliseconds.

Hit Start to enable browser start audio. Wait for sound and filter selection to appear, might take few seconds.

Audio source

Legend:

Sample rate: <waiting audio context>
Audio: <waiting audio context>
Filters:
<waiting audio context>
HP Delay:
0.00 ms0 samp0 cm

Low pass filter:

High pass filter:

"Tweeter" delay (samples):

Integer sample delay (0-96 samples @ 48kHz = 0-2ms)

Info

Delay simulation: At 343 m/s (speed of sound), 1 ms delay = 34.3 cm distance. Delays possible only per sample due to interpolation errors. Typical speaker likely has 0-2cm error, 0-5 samples just due to construction and crossover, and could be more if you listen at an elevated angle..

Notice, this test with headphones isn't equal to what a real speaker in a room would do. For example, loud early reflections (long listening distance, no acoustic treatment) and acoustic summing of the signals affect perceived sound significantly. More over, changing sound is very easy to detect while static "error" might go unnoticed.

If you have DSP speakers you can adjust, try to delay tweeter with your DSP to hear what it does with your system in your room and setup. At short listeing distance it's easy to hear, further out in the room not so much.

Filters are IIR filters whose biquads were exported from VituixCAD with 48kHz sample rate and "Generic" DSP system.

The filters used are listed below:

Credits:

Speech sample:
Demonte, Philippa (2019): Speech corpus - example of edited audio: Harvard_L01_S01_0.wav. University of Salford. Media. https://doi.org/10.17866/rd.salford.7857845.v1
Violin sample:
Pätynen, J., Pulkki, V., and Lokki, T., "Anechoic recording system for symphony orchestra," Acta Acustica united with Acustica, vol. 94, nr. 6, pp. 856-865, November/December 2008. [Online IngentaConnect]
https://research.cs.aalto.fi/acoustics/virtual-acoustics/research/acoustic-measurement-and-analysis/85-anechoic-recordings.html
Music:
Happy Clappy Ukulele by Shane Ivers - https://www.silvermansound.com
Music:
Mirã / Lavoura - MM Moods loop, https://bumpfoot.net/foot276.html