If anything here is confusing, inspiring or absolutely incorrect your comments would be much appreciated.  This is a work in progress and your help improving the information is requested.  Thanks!

< Binaural Recording and Listening

One classic and effective technique for creating stereo is delay.  Significant research has been conducted on the effects of time delay between two speakers.  We already know that we can use delays in the 0 to 1.5 ms range to help Localize a sound between speakers.  As we increase the delay in one speaker beyond 1.5 ms, it starts to add a sense of space, making the sound seem wider.  It spreads out, sounds bigger and remains a single entity.  But it does not tend to sound like a single point source that can be Localized between speakers.  Delay alone, with no difference in volume between channels, creates this lateral (left/right) Spaciousness. 

But if the delayed channel is significantly lower or higher in volume, other effects are heard.  Low enough, the delay becomes inaudible.  When a quiet sound becomes inaudible because a louder sound seems to cover it up, this is called Masking.  Like the sound of a pin drop can be masked by rushing water, a sufficiently quiet delay can become inaudible in the presence of a louder direct sound.  (For more on Masking, see Alex Case’s excellent book Sound FX: Unlocking the Creative Potential of Recording Studio Effects, Chapter 3: Perception.)

But what if the delay is louder than the original?  Even for delays under 25 ms, a louder delayed signal can actually be perceived as a discrete echo.  It no longer sounds like one Spacious image, but two separate sounds, simply because we raised the volume.  Echoes are distinct from the original sound; they create a different feeling than simply seeming wider.  Sometimes you want a single source, sometimes you want a separate echo.  It just depends which one better suits a particular recording.

What’s interesting here is the longer the delay, the easier it is to hear it.  A 10 ms delay that is -20 dB below the original signal will probably be inaudible.  But the same -20dB level is likely to create a sense of Spaciousness at 50 ms.  Similarly a 10 ms delay at 0dB sounds Spacious, whereas at 50ms delay at 0dB sounds like a discrete echo.  A lot of audio folks like to talk about delays under the 35 ms range as the area of Integration, above which delays transition from singular, fused sounds into two separate auditory events.  But volume -- not just the delay time -- is a factor in whether or not the original and delayed sounds will interact for no audible effect, seem like one wide sound or seem like two different sounds.  Volume and time, interrelated, determine if we hear delay and whether that delay seems fused or separate.

Other factors influence the audibility of delays.  Consider the frequency range of the sound.  Broad spectrum signals will generally behave as described above, but narrow range sounds will tend to be less predictable.  Even more significant, the duration of the source will affect the audibility of the delay.  Short, percussive sounds will tend to reveal delay more easily than long, sustained notes.  As with most things audio, the exact numbers don’t matter so much as the patterns:

(1)  Short delay and/or low volume will tend to Mask delay, making it inaudible.

(2)  Longer delay and/or higher volume will tend to reveal delay, which will initially tend to widen the image, and eventually give way to discrete echoes.

(3)  Narrow range signals may not follow the pattern so strictly. 

(4)  Percussive sounds will reveal delay more readily, sustained sounds less readily.

The research about delaying a signal in one speaker relative to another is an incredibly powerful tool for creating stereo.  We can adjust the time and/or volume of a delayed signal to create a fused sense of space, or create a separate echo.  We can even minimize volume and/or reduce the delay time to a point where we can’t hear it.  This may be especially useful if we have several sources of delay but don’t want all of them to be audible in the final stereo presentation.

To some degree the same things hold true for a delayed signal in the same speaker with the original.  Only now the spread of the image is no longer left/right, but instead front/back.  It seems a little easier to Mask the delayed signal in the same speaker.  We will likely need more difference in volume and/or delay time in the same speaker than we are able to accomplish with the delayed signal in a separate speaker.  And that makes sense when you consider that two speakers make better use of The Cocktail Party Effect.  Delay in the same channel with the original can also cause Comb Filtering, which we will examine soon.

Whether you use time difference to delay a signal between two speakers or in the same speaker, this is a powerful concept for creating width and/or depth in stereo.  Placing more than one mic in the same space can create time differences, volume differences, or both.  After placing mics and recording, you can additionally change volume controls and manipulate delay times to minimize delay audibility, widen/deepen a single image, or throw around distinguishable echoes.

Panned Mono >

Audio Wiki front page

Randy Coppinger - who I am and other stuff I'm doing.