Perfect sinusoids/square waves/etc. only exist as mathematical
abstractions. A good starting point would be to get a feel for what, say,
the square wave coming out of an analog synthesizer actually looks like -
the noise floor, the distribution of harmonics, frequency jitter,
under/overshoot, etc.
Your intuition is correct that SIMD can't use the output of one
sub-operation as an input to another in the same vector. This poses a
problem for stuff like recursive filters particularly. For FIR filters it
is not as much of an obstacle, since the output at any particular time is
only a function
For the theoretically inclined: approximating a full Fourier Transform
requires
time interpolation of the samples to a (possibly much higher) sampling
frequency, and on top of that a very long FFT, and proper analysis of the
results of the FFT.
It sounds like you are talking about trying to
If you just want higher frequency resolution, you don't need to do any
oversampling in the time domain. Just zero-pad the time domain signal out
to whatever long length you want, and then use an FFT of that size.
E
On Tue, Mar 31, 2015 at 3:12 PM, MF ukel...@gmail.com wrote:
given a N-point
Yeah if you simply pick out a few peaks you can get the general shape of
the reverb decay, but you miss all of the dense reflections. Multi-tap
delay on its own is fine for the early reflections, but the rest of the
reverb response is more dense as Steffan says.
Keun Sup Lee did some work along
So to you, that Pono player isn't snake oil?
It's more the 192kHz sampling rate that renders the Pono player into snake
oil territory. The extra bits probably aren't getting you much, but the
ridiculous sampling rate can only *hurt* audio quality, while consuming
that much more battery and
.
On 2/10/2015 1:13 PM, Ethan Duni wrote:
I'm all for releasing stuff from improved masters. There's a trend in my
favorite genre (heavy metal) to rerelease a lot of classics in full
dynamic range editions lately. While I'm not sure that all of these
releases really sound much better (how much
bristow-johnson
r...@audioimagination.com wrote:
On 2/10/15 1:51 PM, Ethan Duni wrote:
So to you, that Pono player isn't snake oil?
It's more the 192kHz sampling rate that renders the Pono player into snake
oil territory. The extra bits probably aren't getting you much, but the
ridiculous
How do the crest factors of these different sawtooth waveforms compare?
I'd expect one with randomized phase to have a much lower crest factor.
Which is to say that I'd expect the in-phase sawtooth to activate a lot
more nonlinearity in the playback chain, which explains why that one is
easy to
Thanks for the reference Vicki
What they are hearing is not noise or peaks sitting at the 24th
bit but rather the distortion that goes with truncation at 24b, and
it is said to have a characteristic coloration effect on sound. I'm
aware of an effort to show this with AB/X tests, hopefully it
, as a special case of the general 2 pole biquad
filter.
On 05/02/2015, Ethan Duni ethan.d...@gmail.com wrote:
You just stick the extra zero(s) off at z=infinity.
Does 'z=infinity' mean it's at the origin? I'm not 100% sure of the
terminology used here.
- Peter
--
dupswapdrop -- the music-dsp
P.S. Anyone who knows how to effectively turn ideas into money while
everyone can benefit, let me know. Patenting stuff doesn't sound like
a viable means to me.
Well, that's exactly what patents are for. I'm not sure why you don't
consider that viable. Is it to do with the costs and time required
There is just no way A/B testing on a sample of listeners,
at loud, but still realistic listening levels, would show that
dithering to 16bit makes a difference.
Well, can you refer us to an A/B test that confirms your assertions?
Personally I take a dim view of people telling me that a test would
it takes a little less than a millisecond to receive a 3-byte MIDI
message.
should we put in some kinda delay (like a sample-and-hold) on that which
prevents updating the net control value until 1 or 2 milliseconds after
either
the MSB or LSB is received? or slewing the net control value (slewing
My filter has 2 poles and 1 zero. Unlike the Cookbook filter, which
has 2 poles and 2 zeros.
I think that automatically assumes, the transfer function cannot be
equivalent.
No, that does not follow. A filter with two zeros can produce all of the
transfer functions that a filter with one zero can,
I completely agree! I find it mentally easier to think of energy
stored in each component rather than state variables even though
they are the same. So for musical applications it is important that a
change in the cutoff and resonance doesn't change (until you process
the next sample) the energy
when filters blow up because of varying coefficients, don't they settle
down some finite time after the coefs stop varying?
In a digital implementation? Probably not. Once you have some Inf or
NaN values in your state, you're going to keep getting garbage out
forever unless you explicitly reset
wrote:
On 2/3/15 2:01 PM, Ethan Duni wrote:
I completely agree! I find it mentally easier to think of energy
stored in each component rather than state variables even though
they are the same. So for musical applications it is important that a
change in the cutoff and resonance doesn't change
questions.
E
On Tue, Feb 3, 2015 at 5:13 PM, robert bristow-johnson
r...@audioimagination.com wrote:
On 2/3/15 7:07 PM, Ethan Duni wrote:
well, the output states, y[n-1] and y[n-2], will change if coefs change.
No, those have already been computed and (presumably) output at that
point.
It's
zeros happen to line up with the bin frequencies.
E
On Mon, Dec 8, 2014 at 3:28 PM, Jerry lancebo...@qwest.net wrote:
On Dec 8, 2014, at 2:33 PM, Ethan Duni ethan.d...@gmail.com wrote:
what other presumption is there? i, personally, have never seen a
sequence of
samples of audio
what other presumption is there? i, personally, have never seen a
sequence of
samples of audio or music that was not equidistant and linearly
sampled. it's
what we call uniform sampling.
Some of this new stuff in compressive sensing/sparse reconstruction
involves non-uniform sampling. Not that
, Andreas Tell li...@brainstream-audio.de
wrote:
On 01 Nov 2014, at 02:39, Ethan Duni ethan.d...@gmail.com wrote:
The expectation value of the signal over the ensemble of all unitary
transforms
with a suitable measure (like Haar).
The expected value you describe is equal to the zero signal
...@iki.fi wrote:
On 2014-10-31, Ethan Duni wrote:
Transforms between orthogonal bases are basically rotations. I.e., they
are linear operators that produce each component of the output as a linear
combination of input components. Generally, then, the Central Limit Theorem
tells us
Say we're only taking one length of the FFT transform, and
are only interested in the volume of the various output bins.
Now, how probable is it that we get all equal frequency amounts as
the output of the this FFT transform (without regarding phase), taking
for instance 256 or 4096 bins, and 16
I am not sure if the PDFs are preserved across
transforms from one orthonormal basis to
another, and the answer to your question would
depend on that (Of course it would also depend
on several other parts of the phrasing of your question
that aren't clear to me). My intuition is that PDFs
are
There is a theorem that goes something like this:
If you have white noise expressed in one orthonormal basis, and you
transform it to another orthonormal basis, the result will still be white
noise.
There is certainly no such theorem. For any noise signal you can define
a basis that contains
. You're talking about
something quite different.
E
On Fri, Oct 31, 2014 at 3:42 PM, Andreas Tell li...@brainstream-audio.de
wrote:
On 31 Oct 2014, at 23:31, Ethan Duni ethan.d...@gmail.com wrote:
If you have Gaussian i.i.d. noise, you can apply any unitary transform
you
want and you
On Fri, Oct 31, 2014 at 4:16 PM, Andreas Tell li...@brainstream-audio.de
wrote:
On 01 Nov 2014, at 00:06, Ethan Duni ethan.d...@gmail.com wrote:
The correct statement would be
that an arbitrary unitary transform of a Gaussian white noise signal is
*expected* to give a gaussian white noise
I might find myself in the situation where I am given a nonuniform
distribution. Then Shannon and Hartley formulas would give a different
answer.
And the Hartley formula would be inapplicable, since it assumes a uniform
distribution. So you'd have to use the Shannon framework, both to calculate
Of course it is only a theoretical matter, but we have a
sensible spectrum for our repeating envelope generator,
say at 1 Hertz, with a low pass filter set to 2Hz, possibly
with an actually limited spectrum, that we can multiply
with another wave, which can also be frequency limited
in such a way
Although, it's interesting to me that you might be able to get some
surprising value out of information theory while avoiding any use of
probability ...
Hartley entropy doesn't avoid any use of probability, it simply
introduces the assumption that all probabilities are uniform which greatly
The Hartley entropy
is invariant to the actual distribution (provided all the
probabilities are non-zero, and the sample space remains unchanged).
No, the sample space does not require that any probabilities are nonzero.
It's defined up-front, independently of any probability distribution.
The relevant limit here is:
lim x*log(x) = 0
x-0
It's pretty standard to introduce a convention of 0*log(0) = 0 early on
in information theory texts, since it avoids a lot of messy delta/epsilon
stuff in the later exposition (and since the results cease to make sense
without it, with empty
Sounds like a fun project Scott.
One question though:
Sample rate is approximately 44.6 kHz.
What's with the non-standard sampling rate?
E
On Sun, Oct 12, 2014 at 5:25 PM, Scott Gravenhorst music.ma...@gte.net
wrote:
I've been working on a MIDI Karplus-Strong synthesizer using a Microchip
own is fine,
but your response to attempts to enlighten you is counterproductive. If and
when you get serious about understanding this stuff, this list is here to
help.
E
On Fri, Oct 10, 2014 at 4:21 AM, Peter S peter.schoffhau...@gmail.com
wrote:
n 09/10/2014, Ethan Duni ethan.d...@gmail.com
values in the time
domain is pretty much irrelevant to the question of what its entropy is
(particularly when those values themselves are taken as variable, as you
are doing here).
E
On Thu, Oct 9, 2014 at 9:35 AM, Peter S peter.schoffhau...@gmail.com
wrote:
On 09/10/2014, Ethan Duni ethan.d
to strong conclusions like this. You'll get a lot farther if
you invest a bit into understanding the theoretical framework for this
stuff.
E
On Thu, Oct 9, 2014 at 9:48 AM, Peter S peter.schoffhau...@gmail.com
wrote:
On 09/10/2014, Peter S peter.schoffhau...@gmail.com wrote:
On 09/10/2014, Ethan
- since there is no dependence on previous
samples, there is a constant amount of surprise in each sample,
corresponding to the entropy of the distribution the noise is drawn from.
E
On Thu, Oct 9, 2014 at 10:00 AM, Peter S peter.schoffhau...@gmail.com
wrote:
On 09/10/2014, Ethan Duni ethan.d
capacity of human
audio perception.
E
On Thu, Oct 9, 2014 at 11:11 AM, Peter S peter.schoffhau...@gmail.com
wrote:
On 09/10/2014, Ethan Duni ethan.d...@gmail.com wrote:
Let's assume I have a sinusoidal signal.
Let's assume I amplify it to 10x.
Where does new entropy come from?
It comes from
... since the original question of r b-j was: how can the human ear
convey the high amount of digital PCM information contained on a CD?
Right, my point is that the digital PCM info on a CD typically contains a
*lot* of data that is redundant to human audio perception, and which gets
I did not claim anything about entropy of
continuous signals,
Aren't we talking about impulses in auditory nerves (among other things)?
Those things live in the analog domain.
I was only talking about the entropy
content of digital PCM signals that could be estimated using standard
digital,
Well, your standard options for computing 2 to a fractional power are
either polynomial approximation or table look-up. If I'm reading it
correctly, the approach you quoted there is a second-order polynomial
approximation. You can gain accuracy at the cost of complexity by dialing
up the
Maybe I'm missing something obvious, but shouldn't the filter bank itself
be constant? I.e., no change in the overlap or windowing. The time
stretch/compression is obtained by extrapolating/interpolating the analysis
parameters, not by shifting around the synthesis filter bank relative to
the
to the one exposed here
http://www.ece.uvic.ca/~peterd/48409/Bernardini.pdf, but modified to
work
in real time processing frames of fixed size.
2014-08-19 21:30 GMT+01:00 Ethan Duni ethan.d...@gmail.com:
Maybe I'm missing something obvious, but shouldn't the filter bank itself
be constant
Sinc interpolation would be theoretically correct, but, remember,
that this thread is not about strictily theoretically correct frequency
recognition, but rather about some more intuitive version with the
concept of instant frequency.
What is instant frequency? I have to say that I find this
What are the alternatives to the FFT? Have wavelets been
used for real world solutions?
Sure, wavelets get used. Maybe more in image/video than audio, but I'm
certain someone can come up with some examples of wavelet audio
applications.
If an app needs much higher time resolution
and there are
This means that in principle any piecewise polynomial signal
with bandlimited discontinuities of the signal and its derivatives
is also band limited.
Sorry if I'm missing something obvious, but what is a bandlimited
discontinuity?
E
On Tue, Jul 1, 2014 at 1:17 AM, Vadim Zavalishin
without the minimum phase approach. Even at 44.1 kHz, the lowest
rate you’re likely to make a synth, it would take quite a few samples to
result in enough delay to matter.
On Jul 1, 2014, at 12:29 PM, Ethan Duni ethan.d...@gmail.com wrote:
If you reset the output of a sine wave generator so
rbj
Urs
Regarding the iterative method, unrolling like you did
y0 = y[n-1]
y1 = g * ( x[n] - tanh( y0 ) ) + s
y2 = g * ( x[n] - tanh( y1 ) ) + s
y3 = g * ( x[n] - tanh( y2 ) ) + s
y[n] = y3
is *not* what I described in general.
it *is* precisely equivalent to the example you were
rbj
another semantic to be careful about is transfer function.
we mean something different when it's applied to LTI systems
(the H(z) or H(s)) than when applied to a diode. the latter
semantic i don't use. i would say volt-amp characteristic
of the diode or vacuum tube. or if it was a
It would appear to me that the human hearing system is an LTI system.
It doesn't react in a linear fashion to frequency or loudness, but it
perceives
the same signal the same way at all times, disregarding aging, hearing
loss, etc.
One of the easiest ways to see that hearing must be nonlinear is
the human hearing system is kind of an LTI... only at very low level
processing. The consistency of measured signal (= perceiving the same
signal the same way at all time as somebody wrote here) is present in
the ear canal up to brainstem - inferior colliculus.
My understanding is that there are
So in the digital sense, or in combination with the analog domain,
is it reasonable to think about correctable operations, which as
it were can be inverted, so that applying a digital signal transformation
*and* it's converse, we end up with the same signal or something similar.
I think you mean
So this talk of compressors in the playback chain brings up an important
point. The usual results about CD rate/depth being sufficient are referring
to the signal delivered to the final analog audio output. We all know that
higher rates and depths are appropriate/required for intermediate
Not to be overly antagonistic, but:
I can easily hear the difference between a 192 or 96kHz 24 (or 22 bits +
exponent) bit and downgrading to 48 or 44.1 / 24 bit OR to 192 or 96 kHz 16
bits. Let alone both, easily audible.
If you are hearing obvious differences between those settings, it's a
Hi Doug-
To address some of your general questions about Fourier analysis and
relationship to sampling theory:
Broadly speaking any reasonably well-behaved signal can be decomposed into
a sum of sinusoids (actually complex exponentials but don't worry about
that detail for now). There are
Hi Doug-
Regarding this:
Terms like well behaived when applied to the functon make me wonder
what
stipulations might be implied by the language that you'd have to be a formal
mathmatician to interpret. As an example, I don't even know what the
instrinsic properties of a function may be in this
Yeah this is sometimes called bandpass sampling or under sampling (
http://en.wikipedia.org/wiki/Undersampling) and is commonplace in the
context of like RF communications and such.
But it can also come up in audio applications, for example critically
sampled filter banks. I.e. say you split a
it is, at least, if you accept the EE notion of the Dirac delta function
and not worry so much about it not really being a function, which is
literally what the math folks tell us.
I may be misremembering, but can't non-standard analysis be used to make
that whole Dirac delta function approach
...@audioimagination.com wrote:
On 3/27/14 5:27 PM, Ethan Duni wrote:
it is, at least, if you accept the EE notion of the Dirac delta function
and not worry so much about it not really being a function, which is
literally what the math folks tell us.
I may be misremembering, but can't non
And an AD converter with 2.5 MS/S and 24 bits accuracy sigma delta which
should be great for measurements and undisputably great sampling behavior
for perfect delay lines, etc.:
http://www.analog.com/en/analog-to-digital-converters/ad-converters/ad7760/products/product.html
(eval board:
In the mathematical sense, we could take a Fourier Integral of these
signals neatly added together, and the outcome would be 3 spikes at the
respective frequencies, with the proper amplitude.
No, you get *6* spikes, since these are real-valued signals. So there are
components at both the
Seems to me that you could just reformulate an EM algorithm to work
directly on the actual (magnitude) responses of the biquad stages, and so
bypass the step of converting from a Gaussian response to the actual biquad
response (along with its attendant error). The only obvious wrinkle that
occurs
I recall from the first measurement shown on the analog spectrum analyzer
the source sine wave had 2d or 3d harmonic distortion component of -70 dB.
My power amps are noticeably better than that.
But then they don't have the heterodyne circuits you need for an analogue
frequency analyser
, Theo Verelst theo...@theover.org wrote:
Ethan Duni wrote:
I recall from the first measurement shown on the analog spectrum analyzer
the source sine wave had 2d or 3d harmonic distortion component of -70
dB.
My power amps are noticeably better than that.
But then they don't have
The analog bench gear in that video made me nostalgic... I can almost smell
the solder fumes :]
E
On Thu, Feb 13, 2014 at 6:01 PM, Andrew Simper a...@cytomic.com wrote:
I was really enjoying the proportional Q thread, so in an attempt to
keep in on topic here is a thread where people can
IIRC this is one of those situations where you get to define the value of the
integral at that point however you like (i.e., either the limit from one side
or the other, or a convex combination of both) and all the math then works as
expected under that convention.
Or just point out that it's
I find that the best way to avoid paying for access to scholarly knowledge -
both in legal terms and in effectiveness - is to email the author, express
interest, and ask for any material they can send you. Most times, you'll get
something free, legal, newer and with helpful insight straight
101 - 168 of 168 matches
Mail list logo