funding fundamental frequency(pitch)

Ron N. wrote:

>>>>1. Defining the adequate parametric model of the audio signal.
>>>>2. Defining the cost function to evaluate the parameters of the model.
> 
> 
> You seem to be assuming the existence of closed form
> parametric model for whatever it is that the OP wants
> to measure.

A measurement is worseless if you don't have any idea of what is the 
meaning of this measurement :-) The model should be always.

The troubles with the dumb pitch detectors are because the pitch + 
envelope model is oversimplified. Hence it doesn't matter of how exactly 
do they look for the periodicities: AMDF, autocorrelation, or Mr. 
Teres's black magic. Start with definition of the better model.

> 
>>>>Without (1) and (2), the notion of "pitch" does not make any sense.
> 
> 
> 'course it does.  I've seen little kids can transcribe
> simple melodies, and complain when some nearby players
> are sufficiently sharp or flat.  Are you defining "pitch"
> differently?

I was taught that the engineers never use the words like "sharp", 
"flat", "good", "bad" and such. The adjectives are bullshit, the 
engineers use the numbers.

As for the kids, they detect the pitch by observing the segments of the 
several seconds long, and not in the real time.

> 

>>>Yes, first we should take a spherical horse in vacuum -:)
>>Sure. Why can't we hit the Moon with bow and arrows?
> 
> 
> (1)
> Because the gold standard for the location of the moon
> isn't defined by some impressions formed inside the minds
> of humans (say, trained musicians from a given subculture).

There are the good models for the hearing perception, even standartized 
by ISO. I am pretty sure there are the models for the pitch perception, too.

> (2)
> One probably could from the doorway of an Apollo lunar
> lander.  :)

But why can't I do that from the Earth even with the very best hightech 
bow made of the latest composite materials? :)



Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

On Nov 5, 2:08 pm, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
wrote:
> Ron N. wrote:
> >>>>1. Defining the adequate parametric model of the audio signal.
> >>>>2. Defining the cost function to evaluate the parameters of the model.
>
> > You seem to be assuming the existence of closed form
> > parametric model for whatever it is that the OP wants
> > to measure.
>
> A measurement is worseless if you don't have any idea of what is the
> meaning of this measurement :-) The model should be always.
>
> The troubles with the dumb pitch detectors are because the pitch +
> envelope model is oversimplified.

Agreed.

> Hence it doesn't matter of how exactly
> do they look for the periodicities: AMDF, autocorrelation, or Mr.
> Teres's black magic. Start with definition of the better model.
>
>
>
> >>>>Without (1) and (2), the notion of "pitch" does not make any sense.
>
> > 'course it does.  I've seen little kids can transcribe
> > simple melodies, and complain when some nearby players
> > are sufficiently sharp or flat.  Are you defining "pitch"
> > differently?
>
> I was taught that the engineers never use the words like "sharp",
> "flat", "good", "bad" and such. The adjectives are bullshit, the
> engineers use the numbers.
>
> As for the kids, they detect the pitch by observing the segments of the
> several seconds long, and not in the real time.

The OP did not mention the need for "pitch" measurement
in real time, and in fact seems to be correlating several
adjacent frames (the order of frames wasn't even specified).

Engineers must often design products which get tested by
a random potential customer base, and whether those customers
deem it "good" or "bad" in their opinions, compared to
competitive alternatives available in the same time/cost
frame, may lead to how they vote with their pocketbooks.

> There are the good models for the hearing perception, even standartized
> by ISO. I am pretty sure there are the models for the pitch perception, too.

There do appear to be some research papers.  Are there any
you consider to present a "correct" model for South Indian
music?

Research that I'm interested in finding is whether and
how much human pitch perception might be influenced by the
preceding transient or transitions from previous notes.
The model may have to take into account some time domain
history.


IMHO. YMMV.
--
rhn A.T nicholson d.0.t C-o-M

On Nov 3, 12:37 pm, robert bristow-johnson <r...@audioimagination.com>
wrote:
> On Nov 3, 11:53 am, "mnentwig" <mnent...@elisanet.fi> wrote:
>
> > I have an example program on my webpage that might get you started.
> > However, be aware that audio pitch detection is a science of its own.
>
> >http://www.elisanet.fi/mnentwig/webroot/FFT_peaksearch_audio_example/...
>
> > To give one example, the ear may be tricked into hearing a fundamental
> > that isn't there: If I filter the fundamental away from a piano note,
> > chances are that my ear will still hear it as the original note.
> > There are many older threads on this topic.
>
> i'm still of the opinion that the old AMDF (Average Magnitude
> Difference Function) or a variant (like ASDF with a window and perhaps
> a filter on the difference signal) is the method that makes the fewest
> assumptions.  it only assumes some notion of periodicity and looks for
> the best period, given some error cost weighting applied to the
> difference signal (absolute value and squared are but two possible
> choices).  you look for minimums in that and try to wisely choose (and
> stick with) the right minimum.

The problem I have with using only AMDF is that, while
it is good at finding the lowest period periodicities,
it seems to have poorer high frequency phase (or change of
phase) resolution than do DFT based methods.  An interesting
question is whether the low or high frequency periodicities
more precisely influence subtle differences in perceived pitch.
For instance, pianos are commonly "stretch" tuned, which
more precisely matches the higher harmonics of the low keys
with the high notes, but leaves the low frequency fundamental
periods slightly out-of tune, in terms of the exact ratios.

One method which I prefer is to use FFT based algorithms
to estimate the higher frequency spectral peaks, and then
use something more like ASDF to help determine the "best"
divide down values.

>  that takes a little "expert systems"
> or AIish thinking in the alg.

I think so as well.

> r b-j


IMHO. YMMV.
--
rhn A.T nicholson d.0.t C-o-M

On Nov 6, 6:08 am, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
wrote:
> There are the good models for the hearing perception, even standartized
> by ISO. I am pretty sure there are the models for the pitch perception, too.

Oh, good. I like standards. Does that mean that when our ears diverge
from the standard we need to see a doctor? :-)

Steve

steveu@coppice.org wrote:
> On Nov 6, 6:08 am, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
> wrote:
> 
>>There are the good models for the hearing perception, even standartized
>>by ISO. I am pretty sure there are the models for the pitch perception, too.
> 
> 
> Oh, good. I like standards. Does that mean that when our ears diverge
> from the standard we need to see a doctor? :-)

LOL

Those are the standards, not the regulations. The compliance is 
voluntary, not obligatory :-)

VLV

On Nov 5, 9:31 am, Vladimir Malakhov <schnei...@yandex.ru> wrote:
> Along with "missing fundamental" and "octave problem" I want to add
> one more. This is a problem of simultaneous sounds, when two and more
> strings or two and more instruments play different notes.

yeah, i was deliberately staying in the monophonic pitch detection,
which is a solvable problem.  sorta.

> Can Dmitry
> Terez or someone else recommend any solution? Is situation really
> hopeless?

i dunno.  i seem to remember some AES papers (i think Rob Maher had
one) in source separation.  now while the intended destination of
these were not about pitch detection of multiple sources in a
polyphonic context, i would think the success of polyphonic pitch
detection would be coupled to the success of the source separation
algs.  once you separated the sources, you could apply pitch detection
to each one in parallel.  i dunno.

r b-j

On Nov 5, 4:00 pm, "Ron N." <rhnlo...@yahoo.com> wrote:
> On Nov 5, 9:54 am, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
> wrote:
>
> > Vladimir Malakhov wrote:
> > >>1. Defining the adequate parametric model of the audio signal.
>
> > >>2. Defining the cost function to evaluate the parameters of the model.
>
> You seem to be assuming the existence of closed form
> parametric model for whatever it is that the OP wants
> to measure.
>
> > >>Without (1) and (2), the notion of "pitch" does not make any sense.
>
> 'course it does.  I've seen little kids can transcribe
> simple melodies, and complain when some nearby players
> are sufficiently sharp or flat.  Are you defining "pitch"
> differently?
>
> > >>The above mentioned problems are caused by the poorly defined (1) and (2).
>
> > >>Furthermore, there can't be the ultimate universal pitch detector,
> > >>because it all depends.
>
> > > Yes, first we should take a spherical horse in vacuum -:)
>
> > Sure. Why can't we hit the Moon with bow and arrows?
>
> (1)
> Because the gold standard for the location of the moon
> isn't defined by some impressions formed inside the minds
> of humans (say, trained musicians from a given subculture).
>
> (2)
> One probably could from the doorway of an Apollo lunar
> lander.  :)
>

i just do not know what the fuck this (previous 5 posts) is about?
can someone explain it to me?

r b-j