I'm completing a project that involves synchronising two audio sources over a network, and I need to find some figures that represent what kind delay would be seen as "noticeable".

I've tried searching but haven't found much; all I can find is papers on localisation.

Is anyone aware of any studies or papers that look into this area? Thanks!

  • 1
    $\begingroup$ probably in the order of microseconds...exactly because of source localization. $\endgroup$
    – Memming
    Feb 22, 2014 at 15:56
  • $\begingroup$ Nice question! Welcome to cogsci.SE! $\endgroup$ Feb 22, 2014 at 22:23
  • $\begingroup$ It's not my area, but I imagine some form of search for "signal detection theory" and "audio synchrony" would be a good start (e.g., see this google scholar search). $\endgroup$ Feb 23, 2014 at 23:36
  • 1
    $\begingroup$ You might want to search for "auditory gap detection". $\endgroup$
    – H.Muster
    Feb 24, 2014 at 14:16
  • 1
    $\begingroup$ Another point to consider- delays in audio affect phasing, and if you have sounds arriving at different times to each ear, the brain will interpret that spatially in some instances (i.e. as in trying to localize a sound source) $\endgroup$
    – theMayer
    Feb 28, 2014 at 21:41

1 Answer 1


It greatly depends on what you mean as 'noticeable' - what/why do you want to synchronise, and how it reaches the ears from physical speakers.

Keep in mind that a sound source being 30cm/1 feet further from the ear is about the same effect as a millisecond of delay (speed of sound ~340m/s) - thus, synchronising on the order of microseconds is generally unneccessary unless you somehow have and need sub-millimeter localisation accuracy. That being said, small delays can cause phase cancellation issues which would be noticeable, but would depend on exact placement of sound sources.

However, if we're talking about the limits of the mind - there are two well known phenomena; first is the limit where mind perceives sound as simultaneous with visual stimulus (despite them really being slightly offset), and the second is where mind perceives two close-but-separate sound spikes as a single event (with the loudest masking the weaker). I can't find the exact limits now, but both of them should be somewhere between 1 millisecond and 5 milliseconds if I recall correctly.

  • 1
    $\begingroup$ Interesting subtleties to this question; I wouldn't have guessed. Welcome to cogsci.SE! $\endgroup$ Feb 24, 2014 at 3:54
  • $\begingroup$ I very much doubt that perception of sound-and-video-being-simultaneous goes as low as 5 ms or less. You can try playing video in (for example) VLC and changing the audio offset; you'll see you can go well into the tens of ms before it's clear that something is wrong. This paper mentions professional video editors being able to notice +/- 20msec. I doubt "normals" can go lower. telosalliance.com/images/LA%20White%20Papers/… $\endgroup$
    – hmijail
    Dec 3, 2019 at 7:26
  • $\begingroup$ On the other hand, for delays shorter around 1ms and shorter, one might not only lose the "2-close-but-separate-events", but turn them into spatial location of the sound. $\endgroup$
    – hmijail
    Dec 3, 2019 at 7:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.