Forums

Acoustic memory

Started by Robert Adams May 4, 2008
I accidentally discovered something interesting, and thought I would
share it here.

Prepare a sound file with the following sequence of events.

1) Pink noise, 10 seconds

2) Filtered Pink noise with several notches in the frequency response
(moderate Q's, frequencies in the 200Hz to 5KHz range), 10 seconds

3) Pink noise (no filtering), 10 seconds

You would think that 1) and 3) would sound the same. But what happens
is that your brain adjusts to the notched response in 2), and when you
play 3), you hear pronounced (and very annoying) peaks where the
notches used to be.


This is not one of those "subtle" effects that only audiophiles can
hear; it's such an extreme effect that you are tempted to think the
experiment is rigged somehow.

If you insert a period of silence between 2 and 3, your brain will
eventually reset itself, but it takes more time than you might think.

One thing I would like to try is to insert "reset events" between 2
and 3, other than silence, and see if the "reset period" can be
shortened.

To be honest, I had always thought that people who objected to double-
blind ABX testing were a bit wacky. But now that I understand a bit
more about this memory effect, I must admit that I can see where it
might be difficult to accurately compare two different audio signals.
I also wonder how important frequency response really is, given that
your brain seems to just adapt to the average spectrum anyway.

I know this has appeared in the literature before, but I found it
quite surprising, and it's a very easy experiment to set up yourself.


Bob Adams

"Robert Adams" <robert.adams@analog.com> wrote in message 
news:449e00a8-131c-4f7f-8373-4376ac671f55@c65g2000hsa.googlegroups.com...
>I accidentally discovered something interesting, and thought I would > share it here. > > Prepare a sound file with the following sequence of events. > > 1) Pink noise, 10 seconds > > 2) Filtered Pink noise with several notches in the frequency response > (moderate Q's, frequencies in the 200Hz to 5KHz range), 10 seconds > > 3) Pink noise (no filtering), 10 seconds > > You would think that 1) and 3) would sound the same. But what happens > is that your brain adjusts to the notched response in 2), and when you > play 3), you hear pronounced (and very annoying) peaks where the > notches used to be. > > > This is not one of those "subtle" effects that only audiophiles can > hear; it's such an extreme effect that you are tempted to think the > experiment is rigged somehow. > > If you insert a period of silence between 2 and 3, your brain will > eventually reset itself, but it takes more time than you might think. > > One thing I would like to try is to insert "reset events" between 2 > and 3, other than silence, and see if the "reset period" can be > shortened. > > To be honest, I had always thought that people who objected to double- > blind ABX testing were a bit wacky. But now that I understand a bit > more about this memory effect, I must admit that I can see where it > might be difficult to accurately compare two different audio signals. > I also wonder how important frequency response really is, given that > your brain seems to just adapt to the average spectrum anyway. > > I know this has appeared in the literature before, but I found it > quite surprising, and it's a very easy experiment to set up yourself. > > > Bob Adams
Is this a physical cortical effect or is it the brain doing this sound persistence? hmmm I've been watching a development on this Japanese site, and it shows this 'illusion' for visual stimuli too. See here: http://www.psy.ritsumei.ac.jp/~akitaoka/chcolore.html & http://www.ritsumei.ac.jp/kic/~akitaoka/warp-e.html Somewhere on there, I can't find right now gives a very good example of how size, and more vividly, colour can be affected by surrounding colours and shapes. And we all know we have a persistence of vision, from those 'stare at a picture, then look at a blank wall' games - but I guess some people would find it quite scary that their ears can be fooled too! : ) VC
On May 4, 4:17&#2013266080;am, "VelociChicken" <b...@yahoob.com> wrote:
> "Robert Adams" <robert.ad...@analog.com> wrote in message > > news:449e00a8-131c-4f7f-8373-4376ac671f55@c65g2000hsa.googlegroups.com... > > > > > > >I accidentally discovered something interesting, and thought I would > > share it here. > > > Prepare a sound file with the following sequence of events. > > > 1) Pink noise, 10 seconds > > > 2) Filtered Pink noise with several notches in the frequency response > > (moderate Q's, frequencies in the 200Hz to 5KHz range), 10 seconds > > > 3) Pink noise (no filtering), 10 seconds > > > You would think that 1) and 3) would sound the same. But what happens > > is that your brain adjusts to the notched response in 2), and when you > > play 3), you hear pronounced (and very annoying) peaks where the > > notches used to be. > > > This is not one of those "subtle" effects that only audiophiles can > > hear; it's such an extreme effect that you are tempted to think the > > experiment is rigged somehow. > > > If you insert a period of silence between 2 and 3, your brain will > > eventually reset itself, but it takes more time than you might think. > > > One thing I would like to try is to insert "reset events" between 2 > > and 3, other than silence, and see if the "reset period" can be > > shortened. > > > To be honest, I had always thought that people who objected to double- > > blind ABX testing were a bit wacky. But now that I understand a bit > > more about this memory effect, I must admit that I can see where it > > might be difficult to accurately compare two different audio signals. > > I also wonder how important frequency response really is, given that > > your brain seems to just adapt to the average spectrum anyway. > > > I know this has appeared in the literature before, but I found it > > quite surprising, and it's a very easy experiment to set up yourself. > > > Bob Adams > > Is this a physical cortical effect or is it the brain doing this sound > persistence? hmmm > > I've been watching a development on this Japanese site, and it shows this > 'illusion' for visual stimuli too. > See here: &#2013266080;http://www.psy.ritsumei.ac.jp/~akitaoka/chcolore.html&#2013266080;&http://www.ritsumei.ac.jp/kic/~akitaoka/warp-e.html > Somewhere on there, I can't find right now gives a very good example of how > size, and more vividly, colour can be affected by surrounding &#2013266080;colours and > shapes. > > And we all know we have a persistence of vision, from those 'stare at a > picture, then look at a blank wall' games - but I guess some people would > find it quite scary that their ears can be fooled too! &#2013266080;: ) > > VC- Hide quoted text - > > - Show quoted text -
crossposted interesting thread to rec.audio.pro Mark
On 4 Mai, 05:03, Robert Adams <robert.ad...@analog.com> wrote:

> To be honest, I had always thought that people who objected to double- > blind ABX testing were a bit wacky. But now that I understand a bit > more about this memory effect, I must admit that I can see where it > might be difficult to accurately compare two different audio signals. > I also wonder how important frequency response really is, given that > your brain seems to just adapt to the average spectrum anyway. > > I know this has appeared in the literature before, but I found it > quite surprising, and it's a very easy experiment to set up yourself.
These kinds of things serve as very necessary reminders never to take anything for granted, nor assume anything to be easy or straight-forward whenever human sensory systems are involved. Thanks for an intriguing experiment. Rune