### robert bristow-johnson (@rbj)

Thanks for the link, Rick. I have done the Rader/Gold quadrature oscillator myself with the 56K back around 1992. I had to have the fixed-point coefficients...

>"Also, different initial conditions (what you call y[n-1] and y[n-2]) didn’t seem to have any discernable effect on the output of my resonator."Of course,...

Did you do this?: y[n] = (2 cos(Ω))y[n-1] - y[n-2]initial states (given the parameters, A, Ω, φ): y[-1] = A cos(-Ω + φ) y[-2] = A cos(-2Ω + φ)And...

So then, I would expect that, even if the resonant sinusoid from the noise is at some random (and slowly changing) phase, it would team up with the intended sinusoid...

If it's additive, it's a source. Use superposition. But, of course, modeling quantization error as additive noise is an assumption. With the non-linear reality,...

"The effect of quantized coefficients is that the oscillator's poles stay on the unit circle but their angles become shifted from the desired angle."I understand...

About a quarter century ago, I was schooled a little by James McCartney (creator of SuperCollider) about this simple sinusoidal oscillator: y[n] = (2 cos(Ω))y[n-1]...

Kewl!I recognize your "73" from my ham radio days 4 decades ago. (I used to be WB∅CCA.)So have you made little radio widgets with an ARM processor? From bare...

Long time ago (45 years), I took a class in antenna design and it was like an analog signal processing class. Like using Tchebyshev polynomials to design antenna...

I had explored this title before and I could not make it work as promised. It seemed to me to be a little "something for nothing" trick, which also means "If it...

It looks like you're reviving a TI application. But if you're doing filtering in fixed-point with a general-purpose CPU (that doesn't do floating point), I have...

It's the denominator coefficients, a1 and a2, that are getting quantized.The specific frequency might be off, due to quantizing a1, but the radii of the pair of...

There aren't really any DACs that output a sinc function for each sample. That would be really ideal (in a theoretical sense) but it's a mathematical model for...

They do. I would have to find the right textbook or paper. Maybe Julius Smith has it spelled out. Look at https://ccrma.stanford.edu/~jos/fp/Converting_Stat...BTW,...

The state space model (like we get when we take a Control Systems course) is much more general. I.e. the very same transfer function can be implemented multiple...

But the same can be said for the Short-Time Fourier Transform (STFT).But, as best as I can tell, the analysis window of a wavelet transform gets smaller for higher...

Well you're doing a mpysp() operation (and a complex addition, which doesn't cost much) 280,000 times. Why does one run cost 192 instruction cycles?And is this...

Dean,In the digital signal world, the waveform is made up of numbers. A stream of numbers called "samples". There isn't an "energy" in the sense that you have...

this is what i meant to say, but when i hit "Submit Reply", the response was something like "You must be logged in to reply" and it totally lost about 5 or 6 paragraphs.as...

i really dunno diddley, Tim, but isn't the DCT (there are four or five forms) essentially a DFT of 4 times the size (twice in X and twice in Y) where the added...

Not sure your company wants to consider it but Analog Devices has a relatively new fire-breathing SoC that has an ARM and two SHArC DSPs on one chip. But I dunno...

By "vector", I presume you mean a finite sequence $\{ v[n] \} $ of $ N $ samples of a signal and RMS is $$ V_\mathrm{rms} \triangleq \sqrt{\frac{1}{N} \sum\limits_{n=0}^{N-1}...

I can send you some C code that does wavetable synthesis with linear interpolation between samples (this is good enough if the wavetables are large enough that the...

Okay, if the waveform is a quasi-periodic note or tone, then all of the "partials" or overtones are virtually harmonic, i.e. every sinusoid (or partial) has...

why not just do Wavetable Synthesis? you can IFFT and scale the waveforms in advance. is there a reason you must do it real time on the FPGA?

so you LPF it, like you might with PWM, right? are we talking about the output of a sigma-delta modulator?

sure i could be interested. it would have to be audio DSP oriented. effects or synthesis.but i am not really working now, so i would have to dust something off. ...

You are absolutely correct. I was reverberating what I read at https://ccrma.stanford.edu/~jos/Interpolation/Farrow_Interpolation_Features.html and I didn't drill...

you need to zero-pad your x signal before the FFT. delaying the signal by linear phase-shifting the spectrum is exactly an example of using the FFT to do what...

i dunno how that's gonna do it. blind de-reverb removal requires some kind of automatic learning regarding your input-output relationship. the effect of reverberation...

i remember seeing such papers in JAES.had something to do with recognizing later replicas of earlier features. i think from that you might be able to infer possible...

yes, the term "pass" means processing the entire FFT array once. if \(N = 2^p\) is the size of the FFT, then \( p = \log_2(N) \) is the number passes of the radix-2...

Stephane, it's too late i know, but when you were implementing mathjax or whatever mechanism to render \(\LaTeX\) math, i would have picked an existing protocol...

You actually want two bits of headroom, not just one. you could have maxxed both your real and imaginary parts, and then have that turned 45 degrees by the FFT...

If you wanna design this block-floating-point FFT from scratch, I would recommend that you find yourself a manual for the old Motorola DSP56002 or later (the 56000...

In the audio industry, a sorta consumer-oriented industry that is pretty close to the market (like nearly all of the time I was working on development of a product...

Stephane, if you can see the edit history of this post, i tried it all ways including with the double $. it never works when i do it.

maybe you should just credit Jim Shima for the math that is in his own thesis.but it's not uncommonly known that the difference in angles or arguments is$$ \arg...

so there are 26 letters and 10 numerals. i remember old typewriters had "l" for "1" and "O" for "0". but that only gets you down to 34. what other symbols...

never learned Baudot-Murray nor Reed-Solomon. more familiar with Linkwitz-Riley or Sallen-Key or Karplus-Strong.i know a little about error detection, but little...

kaz, even doing a butterfly is implementation dependent, i think, because how you gotta handle the indexing of the two in-place bins and then doing the add/subtract...

So you "have a quadrature FM signal that is modulated on a low frequency IF" which sounds to me that you have both \$i[n]\$ and \$q[n]\$ of a complex signal that...

Okay, do you know what "Convolutional Reverb" is? What "Fast Convolution" is?Do you know what the Schroeder Reverb model is or the Jot Reverb model? There might...

it turns out that *every* digital filter is a comb filter (of sorts). if the digital filter is a LPF, you will see comb teeth at every integer multiple of the...

i do not see why the Doppler shift effect should be any different for the modulation than it is for the carrier. the whole signal gets scaled in time by a single...

okay, i understand this statement: "I have used a Farrow interpolator to do fractional rate sample rate conversion on only one of the signals..." but i am curious...

if you're interpolating (or sample rate converting) the stretched replica to make it the same length as the original signal segment, how do you know what the sample...

well, i don't have the fixed=point toolbox, but if you're plagued with limit cycles due to round-to-nearest quantization, my suggestion is that you use a technique...

But then you have to toss in the modulo indexing in order to use the theorems in your "(i.e., discrete Fourier transform)". That modulo indexing makes it periodic.Uniform...

Stephan, your implementation of \( \LaTeX \) here leaves a bit to be desired. It works erratically. especially when i edit a post that had math previously appearing.if...

Well, this is an old issue with me and Eric that dates back to the mid 1990s on comp.dsp . I must differ with Eric where he writes:"There do not need to be any...

I don't see any truncation or cast to a fixed-point value or any nonlinearity. So I don't understand what the source of any limit cycle would be.At least, in my...

The very basics of the LMS adaptive filter are not all that difficult to understand. Perhaps ask a question about it at the Signal Processing Stack Exchange and...

yeah, what John says. because reducing 1024 degrees of freedom down to 3 degrees of freedom might not so good. even if your FIR taps are not adjacent to each...

i do now. but it looks a little Maclaurenish to me. since, ultimately, the power series will have to be finite (which is a polynomial), you will want (slightly) different...

but can't you develop a Cordic operation based on 2^x instead of e^x just as efficiently? there might be an internal scaling operation in there that would be different...

it just seems to me, Brian, that especially for a fixed-point implementation, that 2^x is the simplest primitive and then use A^x = 2^(x*log2(A)).When it's base...

In this stackexchange answer are the coefficients for some short finite-order polynomials for evaluating (at a pretty good accuracy) a bunch of transcendental functions...

well, it appears to me that you will need to perform Asynchronous Sample Rate Conversion (ASRC). there are chips to do that, but if you have the sufficient unused...

Okay, so I know the SHArC core instruction set pretty well and some of the key features that existed in the SHArC back in the olden daze (like 1995 with version...

Do you have an ICE-1000 or some JTAG emulator with this EZKIT?If you don't use a JTAG emulator, I s'pose ADI provides some means of loading and running code, but...

"If the DTFT of a sequence has a bandwidth omega_max > pi/2, a downsampling by a factor of 2 is surely going to cause aliasing in the resulting DTFT."Yes it will. ...

Similarly to some filterbank and wavelet transforms, there is perfect reconstruction because the aliases have a way of killing each other off.But when thinking about...

I still have yet to really do any work on an ARM even though I am told that some C code I wrote (and tested with Xcode) found its application on a phone running...

I am only responding to the title of the post.It seems to me that not very many books that describe fast convolution consider the optimal size for the FFT and most...

There is more to the various "analog effects" than filters and convolution. If you're trying to simulate a guitar amplifier with different cabinets, perhaps the...

no. it's an alternative way of doing it. using polyphase FIR, you have coefficients for an FIR filter for every fractional delay you want.Let's say your FIR...

i have never heard of the "Itakura-saito distance" before. i did know about the Log-Spectral distance which is in dB.the form of this the integrand is (x-1) -...

I dunno. Sometimes mathematics is useful for economic analysis and projections that should influence politics. It's politically useful to understand a runaway...

Not only is knowledge in DSP **very** useful designing digital controllers, but Control Theory has some concepts that are useful for DSPers that never got paid...

So my understanding is that you want a fractional-sample delay, where you specify in advance how much delay you want.1. If the delay is a fraction of a sample less...

yeah, it was Sokolich (who was around the alt.sci.physics.acoustics newsgroup when it was active) not the goofy brit named Gareth.didn't remember that i was scraping...

well, i never much hung around these specific parts. this might be my very first DSPrelated post.i remember Bob Cain from the comp.dsp daze. if i remember right,...

Use this form to contact **rbj**

Before you can contact a member of the *Related Sites:

- You must be logged in (register here)
- You must confirm you email address