keeping the beat

Richard Dobson <richarddobson@blueyonder.co.uk> wrote:
> On 28/08/2011 22:11, Vladimir Vassilevsky wrote:
> ..
> >> Is there any theory regarding what's happening, at the
> >> neural level? How does the ear filter the beat?
> >
> > Detecting the beat is a trivial DSP task. It has nothing to do with DFT
> > or the fundamental frequency.
> >
> >> It might be hard to separate this specific question
> >> from the whole huge puzzle of how the brain
> >> processes music.
> >
> > No need for brains, a mere PIC would do.
> >
> Strange, then, that so many beat detector tools can get it very
> amusingly wrong. They invariably rely on certain basic assumptions about
> the genre being tracked. Confusing duple and quadruple time is merely
> the most common error. ~If~ you assume the beat is indicated by a single
> equally spaced transient "on the beat", you may well be able to do it
> with "a mere PIC". The reality is usually rather different, and is,
> among other factors, culturally determined at least as much as it is
> acoustically determined. So I am led to wonder, whether you have
> yourself actually made a beat detector? And if so, which genres of music
> did you test it on?

I'm sure it worked just fine on these 3 disco songs they tested it on!

> You might like to consider certain styles of African drumming. It was
> explained to me once this way - "everyone knows where the beat is, so
> nobody actually needs to play it".

:)

> Richard Dobson

I don't know much about beat detection or music, but your post wins as
"best of thread" in my book! :)

The brain is a very complex temporal pattern detector (and generator).
Though a lot of pop music has very simple underlying drum or base lines
where the beat can be easily extracted, the brain is in no way limited to
such simplicity, as we find with the more complex forms of music that exist
- seemingly favored by people with the greater training and exposure to
music.  Ultimately, the rhythm patterns are culturally defined, and if you
have't programmed the full culture into the DSP (as it gets "programmed"
into the brain of a trained musician), then beat detection is likely to
fail at times.

-- 
Curt Welch                                            http://CurtWelch.Com/
curt@kcwc.com                                        http://NewsReader.Com/

On Aug 29, 2:38&#4294967295;am, Richard Dobson <richarddob...@blueyonder.co.uk>
wrote:
>  "everyone knows where the beat is, so nobody actually needs to play it".

Awesome quote!  And it takes me back to the days that I took a class
in African drumming.  The quote is pretty true, though I would say
that the downbeat always gets plenty of emphasis.

"RichD" <r_delaney2001@yahoo.com> wrote in message 
news:64ba3fe6-8011-42f5-a8a8-8b9b8cf1ae91@e20g2000prn.googlegroups.com...
> Last week I attended a blues concert.  It was called
> blues, but they rocked!  Not the traditional ballad style,
> but high energy.
>
> At one point, during an instrumental, the audience clapped
> along with the music.  The band included electric guitar
> and bass, acoustic bass, piano, drums, and trumpet.
> It was not a simple melody, the orchestration was complex,
> each instrument playing its own part.
>
> During this passage, I started wondering, how do the
> listeners so easily follow the beat?  If you're a
> mathematican, you'd probably do DFT, and pick out
> the fundamental.  I find it hard to imagine the brain
> does this.
>
> Is there any theory regarding what's happening, at the
> neural level?  How does the ear filter the beat?
>
> It might be hard to separate this specific question
> from the whole huge puzzle of how the brain
> processes music.

Musicians train with metronomes and "counting" as they call it.

Because of the regularity, they can detect changes easily.

Their performance may well be off beat and more musical accordingly.

I have two daughters who are serious musicians (classical and jazz) and they can both play 
by (a) reading the music, plus (b) following the tempo from what they hear, plus (c) 
reading the conductors eyes.

How to track the "beat"? Fourier space wouldn't pick up changes fast enough IMO. A combo 
of Fourier and predictor/corrector would probably be close to human capability. Determine 
the typical envelope that finds the beat, but pay more attention to changes. BTW, drummers 
are pretty consistent, but miss beats.

Tom. 

RichD wrote:
> Last week I attended a blues concert.  It was called
> blues, but they rocked!  Not the traditional ballad style,
> but high energy.
>
> At one point, during an instrumental, the audience clapped
> along with the music.  The band included electric guitar
> and bass, acoustic bass, piano, drums, and trumpet.
> It was not a simple melody, the orchestration was complex,
> each instrument playing its own part.
>
> During this passage, I started wondering, how do the
> listeners so easily follow the beat?

The short answer is not all do. I speak from personal experience ;)
My musically trained friends consider my inaility to follow a 
beat/rhythm comical.

For example, back in the 60's International Folk Dancing was popular 
on college campuses. There were primarily three genres popular with 
our group - Western European (primarily Germanic or British Isles), 
South Eastern European (south of Austria, north of Turkey), and 
Israeli. I could follow the Western European or Israeli music. The 
other -- not well at all. A friend with a master's in Music 
Education tried many hays to help me pick out the beat/rhythm of one 
dance unsuccessfully. One tool was to notice the similarity of the 
dance rhythm to that of "mixed pickles". I never got it. But the 7 
and 8 year olds she taught in public school had no problem with it.

Others have pointed out the importance of the musical tradition of 
the listener. I'd ask any linguists in the group if there is any 
correlation of the"rhythm" of the language of an area and its music?

How is the mechanics of the hearing process involved? I've had a 
severe high frequency loss in one ear since a child due to a series 
of ear infections. In my 30's series of tests determined it was 
"nerve damage". Why is "nerve damage" so frequency sensitive - (most 
plots of my hearing in that ear look like a picket fence)?

Another question is how are physical coordination skills related to 
learning to recognize rhythm. I had a brain injury during birth that 
left me paralyzed on one side for several months with effects my 
mother could spot lasting well into my adult years.

A fascinating topic.



> If you're a
> mathematican, you'd probably do DFT, and pick out
> the fundamental.  I find it hard to imagine the brain
> does this.
>
> Is there any theory regarding what's happening, at the
> neural level?  How does the ear filter the beat?
>
> It might be hard to separate this specific question
> from the whole huge puzzle of how the brain
> processes music.
>
> --
> Rich

On Tue, 30 Aug 2011 10:27:22 -0500, Richard Owlett
<rowlett@pcnetinc.com> wrote:

>RichD wrote:
>> Last week I attended a blues concert.  It was called
>> blues, but they rocked!  Not the traditional ballad style,
>> but high energy.
>>
>> At one point, during an instrumental, the audience clapped
>> along with the music.  The band included electric guitar
>> and bass, acoustic bass, piano, drums, and trumpet.
>> It was not a simple melody, the orchestration was complex,
>> each instrument playing its own part.
>>
>> During this passage, I started wondering, how do the
>> listeners so easily follow the beat?
>
>The short answer is not all do. I speak from personal experience ;)
>My musically trained friends consider my inaility to follow a 
>beat/rhythm comical.
>
>For example, back in the 60's International Folk Dancing was popular 
>on college campuses. There were primarily three genres popular with 
>our group - Western European (primarily Germanic or British Isles), 
>South Eastern European (south of Austria, north of Turkey), and 
>Israeli. I could follow the Western European or Israeli music. The 
>other -- not well at all. A friend with a master's in Music 
>Education tried many hays to help me pick out the beat/rhythm of one 
>dance unsuccessfully. One tool was to notice the similarity of the 
>dance rhythm to that of "mixed pickles". I never got it. But the 7 
>and 8 year olds she taught in public school had no problem with it.
>
>Others have pointed out the importance of the musical tradition of 
>the listener. I'd ask any linguists in the group if there is any 
>correlation of the"rhythm" of the language of an area and its music?
>
>How is the mechanics of the hearing process involved? I've had a 
>severe high frequency loss in one ear since a child due to a series 
>of ear infections. In my 30's series of tests determined it was 
>"nerve damage". Why is "nerve damage" so frequency sensitive - (most 
>plots of my hearing in that ear look like a picket fence)?

There are groups of hairs arranged along the inside of the cochlea
that essentially act as narrow bandpass filters.   Each stimulate a
nerve so that the brain gets a picture of the input stimulus with
frequency isolation.   When a particular set of hairs gets destroyed
or their nerve goes inop, then you lose that frequency range of your
hearing.

What happens sometimes is that the hairs get broken off, and they
don't regrow in humans (although they do in other animals, like
sheep).   My understanding is that this can result in either loss of
hearing at those frequencies or tinnitus at that frequency.

I learned some of this when I developed tinnitus about fifteen years
ago.


>Another question is how are physical coordination skills related to 
>learning to recognize rhythm. I had a brain injury during birth that 
>left me paralyzed on one side for several months with effects my 
>mother could spot lasting well into my adult years.
>
>A fascinating topic.
>
>
>
>> If you're a
>> mathematican, you'd probably do DFT, and pick out
>> the fundamental.  I find it hard to imagine the brain
>> does this.
>>
>> Is there any theory regarding what's happening, at the
>> neural level?  How does the ear filter the beat?
>>
>> It might be hard to separate this specific question
>> from the whole huge puzzle of how the brain
>> processes music.
>>
>> --
>> Rich
>

Eric Jacobsen
http://www.ericjacobsen.org
http://www.dsprelated.com/blogs-1//Eric_Jacobsen.php

Eric Jacobsen wrote:
> On Tue, 30 Aug 2011 10:27:22 -0500, Richard Owlett
> <rowlett@pcnetinc.com>  wrote:
>
>> RichD wrote:
>>> Last week I attended a blues concert.  It was called
>>> blues, but they rocked!  Not the traditional ballad style,
>>> but high energy.
>>>
>>> At one point, during an instrumental, the audience clapped
>>> along with the music.  The band included electric guitar
>>> and bass, acoustic bass, piano, drums, and trumpet.
>>> It was not a simple melody, the orchestration was complex,
>>> each instrument playing its own part.
>>>
>>> During this passage, I started wondering, how do the
>>> listeners so easily follow the beat?
>>
>> The short answer is not all do. I speak from personal experience ;)
>> My musically trained friends consider my inaility to follow a
>> beat/rhythm comical.
>>
>> For example, back in the 60's International Folk Dancing was popular
>> on college campuses. There were primarily three genres popular with
>> our group - Western European (primarily Germanic or British Isles),
>> South Eastern European (south of Austria, north of Turkey), and
>> Israeli. I could follow the Western European or Israeli music. The
>> other -- not well at all. A friend with a master's in Music
>> Education tried many hays to help me pick out the beat/rhythm of one
>> dance unsuccessfully. One tool was to notice the similarity of the
>> dance rhythm to that of "mixed pickles". I never got it. But the 7
>> and 8 year olds she taught in public school had no problem with it.
>>
>> Others have pointed out the importance of the musical tradition of
>> the listener. I'd ask any linguists in the group if there is any
>> correlation of the"rhythm" of the language of an area and its music?
>>
>> How is the mechanics of the hearing process involved? I've had a
>> severe high frequency loss in one ear since a child due to a series
>> of ear infections. In my 30's series of tests determined it was
>> "nerve damage". Why is "nerve damage" so frequency sensitive - (most
>> plots of my hearing in that ear look like a picket fence)?
>
> There are groups of hairs arranged along the inside of the cochlea
> that essentially act as narrow bandpass filters.   Each stimulate a
> nerve so that the brain gets a picture of the input stimulus with
> frequency isolation.   When a particular set of hairs gets destroyed
> or their nerve goes inop, then you lose that frequency range of your
> hearing.
>
> What happens sometimes is that the hairs get broken off, and they
> don't regrow in humans (although they do in other animals, like
> sheep).   My understanding is that this can result in either loss of
> hearing at those frequencies or tinnitus at that frequency.
>
> I learned some of this when I developed tinnitus about fifteen years
> ago.
>

Very interesting. Thank you.
I doubt the testing method that led to the "nerve deafness" 
diagnosis could have distinguished broken hairs from nerve damage. 
It basically compared response via traditional headphones and 
stimulus via bone conduction. I've had tinnitus (noticeable in my 
"good" ear) since childhood. Never understood why people referred to 
it as "ringing in the ear". I always thought of it as a high pitched 
whistle. As my eardrum(s?) were surgically punctured at least twice 
to drain pus, I could see the hairs being mechanically damaged. When 
I got to college and learned of "transfer functions" I pictured my 
strange hearing response as possibly due to resonances due to 
scarring of the eardrum. Broken hairs would conform to physics better.

Wonder if the tinnitus appearing in "good" ear could be explained by 
both ears receiving a noise stimulus and only the "good" ear being 
able to respond to it?

Any readable competent articles on the web meant for lay persons you 
would recommend? I have a related interest in why computer speech 
recognition problems don't seem to map to problems with human speech 
recognition.





>
>> Another question is how are physical coordination skills related to
>> learning to recognize rhythm. I had a brain injury during birth that
>> left me paralyzed on one side for several months with effects my
>> mother could spot lasting well into my adult years.
>>
>> A fascinating topic.
>>
>>
>>
>>> If you're a
>>> mathematican, you'd probably do DFT, and pick out
>>> the fundamental.  I find it hard to imagine the brain
>>> does this.
>>>
>>> Is there any theory regarding what's happening, at the
>>> neural level?  How does the ear filter the beat?
>>>
>>> It might be hard to separate this specific question
>>> from the whole huge puzzle of how the brain
>>> processes music.
>>>
>>> --
>>> Rich
>>
>
> Eric Jacobsen
> http://www.ericjacobsen.org
> http://www.dsprelated.com/blogs-1//Eric_Jacobsen.php

On Tue, 30 Aug 2011 03:08:00 -0700 (PDT), John Ladasky
<ladasky@my-deja.com> wrote:

>On Aug 29, 2:38&#4294967295;am, Richard Dobson <richarddob...@blueyonder.co.uk>
>wrote:
>>  "everyone knows where the beat is, so nobody actually needs to play it".
>
>Awesome quote!  And it takes me back to the days that I took a class
>in African drumming.  The quote is pretty true, though I would say
>that the downbeat always gets plenty of emphasis.



How about automatically detecting the beat of say,  an acapella group
or  barbershop quartet ??

boB

On Aug 28, Vladimir Vassilevsky <nos...@nowhere.com> wrote:
> > Vlad, how is it that you are so willing to spout off
> > something with such a large confidence-ignorance product?
>
> >>Detecting the beat is a trivial DSP task. It has nothing to
> >>do with DFT or the fundamental frequency.
>
> > perhaps google "Jean Laroche beat detection"
> > without quotes and see what you get. &#4294967295;it's so easy to
> > check the veracity of what some of your neurons are
> > firing that you might want to do that
> > before committing it to text and broadcasting it.
>
> Detecting a beat is trivial and primitive DSP task. Even
> a stupident or matlabi could do that.

ok, Boy Wonder, clue us in how to detect the
beat without spectral decomposition.

--
Rich

On Aug 28, 1:38&#4294967295;pm, RichD <r_delaney2...@yahoo.com> wrote:
> Last week I attended a blues concert. &#4294967295;It was called
> blues, but they rocked! &#4294967295;Not the traditional ballad style,
> but high energy.
>
> At one point, during an instrumental, the audience clapped
> along with the music. &#4294967295;The band included electric guitar
> and bass, acoustic bass, piano, drums, and trumpet.
> It was not a simple melody, the orchestration was complex,
> each instrument playing its own part.
>
> During this passage, I started wondering, how do the
> listeners so easily follow the beat? &#4294967295;If you're a
> mathematican, you'd probably do DFT, and pick out
> the fundamental. &#4294967295;I find it hard to imagine the brain
> does this.
>
> Is there any theory regarding what's happening, at the
> neural level? &#4294967295;How does the ear filter the beat?
>
> It might be hard to separate this specific question
> from the whole huge puzzle of how the brain
> processes music.

My limited understanding is that the ear-brain has pattern
recognizers and something like delay lines.  It will find
patterns from memory and then use the delay lines to line
up those patterns (even if the beats aren't even actually
there in the sound, but only implied by the best fit patterns).

A PIC DSP likely can't do this if the beats are implied
by the music, but not actually present in the sound.
(e.g. leave out the down beats from a common tune,
and the typical kid can probably figure out where the
missing notes should have been, can clap to those).


IMHO. YMMV.
--
rhn A.T nicholson d.0.t com
http://www.nicholson.com/rhn/

On Aug 30, 9:57&#4294967295;pm, RichD <r_delaney2...@yahoo.com> wrote:
> On Aug 28, Vladimir Vassilevsky <nos...@nowhere.com> wrote:
>
> > > Vlad, how is it that you are so willing to spout off
> > > something with such a large confidence-ignorance product?
>
> > >>Detecting the beat is a trivial DSP task. It has nothing to
> > >>do with DFT or the fundamental frequency.
>
> > > perhaps google "Jean Laroche beat detection"
> > > without quotes and see what you get. &#4294967295;it's so easy to
> > > check the veracity of what some of your neurons are
> > > firing that you might want to do that
> > > before committing it to text and broadcasting it.
>
> > Detecting a beat is trivial and primitive DSP task. Even
> > a stupident or matlabi could do that.
>
> ok, Boy Wonder, clue us in how to detect the
> beat without spectral decomposition.
>
> --
> Rich

Autocorrelation surely