partitioned/fast convolution

Started by peterworth in comp.dsp11 years ago 22 replies

i'm trying to implement a real-time partitioned/fast convolution and having real trouble getting my head around which partitions to multiply...

i'm trying to implement a real-time partitioned/fast convolution and having real trouble getting my head around which partitions to multiply with which, and what to cumulatively sum and output etc... can anyone offer or link to some pseudocode explaining it step by step or a diagram? thanks.


Convolution of symmetric data

Started by mpb in comp.dsp10 years ago 18 replies

Hi All! I have to convolute two sets of symmetric data which are zero centered i.e. c=a*b where for example a=[1 2 3 4 5 4 3 2 1] b=[2...

Hi All! I have to convolute two sets of symmetric data which are zero centered i.e. c=a*b where for example a=[1 2 3 4 5 4 3 2 1] b=[2 4 6 8 10 8 6 4 2] The data may be also odd a=[5 4 3 2 0 -2 -3 -4 -5] etc. etc. I can perform the convolution either by FFT or by direct multiplication. But I want to make use of the odd/even symmetry of the data. If I want to use FFTW, it requi...


Convolution of 2 different sized arrays of data?

Started by Fleabag in comp.dsp13 years ago 18 replies

Hi I am not totally new to DSP, I did do a module in it at University as part of my Masters, but that was around 15 years ago... I am...

Hi I am not totally new to DSP, I did do a module in it at University as part of my Masters, but that was around 15 years ago... I am looking at implementing audio effects, such as reverberation and amplifier/speaker modelling, by the convolution of an audio signal with an impulse response recorded from a room or amplifier. Specifically I want to implement a bass guitar amp model to improve ...


Cyclic Prefix in OFDM System

Started by runinrainy in comp.dsp5 years ago 1 reply

Could anyone please explain how Cyclic Prefix in OFDM system converts linear convolution to the circular convolution? Thanks in advance! ...

Could anyone please explain how Cyclic Prefix in OFDM system converts linear convolution to the circular convolution? Thanks in advance! _____________________________ Posted through www.DSPRelated.com


fftw3 roundoff error in convolution c++

Started by Kwisatch in comp.dsp10 years ago 7 replies

Hi I'm using fftw3 on ubuntu linux in a convolution of 1-dimensional data. for ( int k=0 ; k

Hi I'm using fftw3 on ubuntu linux in a convolution of 1-dimensional data. for ( int k=0 ; k


fftw3 roundoff error in convolution c++

Started by Kwisatch in comp.dsp9 years ago 6 replies

Hi I'm using fftw3 on ubuntu linux in a convolution of 1-dimensional data. for ( int k=0 ; k

Hi I'm using fftw3 on ubuntu linux in a convolution of 1-dimensional data. for ( int k=0 ; k


filtering by convolution

Started by RMcPherson in comp.dsp10 years ago 1 reply

Hi all, I have some newbie questions for someone much more experienced than I. I am implementing a 6 band equalizer on the AVR32 processor,...

Hi all, I have some newbie questions for someone much more experienced than I. I am implementing a 6 band equalizer on the AVR32 processor, which has built in functions that do filtering by convolution of the input signal with the impulse response of filter. I have a couple of questions based on the AVR32 implementation of FIR filtering. In matlab I generated 6 bandpass filters and a hig...


Another Paper - maybe slightly OT - Euler's Totient

Started by Anonymous in comp.dsp10 years ago 3 replies

Hello All, I've put together another paper this time detailing some topics in number theory and public key encryption. One topic covered is...

Hello All, I've put together another paper this time detailing some topics in number theory and public key encryption. One topic covered is Dirichlet convolution which is different from the convolution we "normally" encounter in DSP and engineering. So I think you may find this interesting. http://www.claysturner.com/dsp/totient.pdf As usual I appreciate any and all feedback. Clay ...


True stereo convolution

Started by VelociChicken in comp.dsp11 years ago 10 replies

Hello, I've written my own convolution algorithm using FFT, all seems to work nicely and it's fairly fast. But I've got a little stuck on making...

Hello, I've written my own convolution algorithm using FFT, all seems to work nicely and it's fairly fast. But I've got a little stuck on making it true stereo. Originally I thought I just simply use two convolutions, but that does not work because the impulses fed into the system work on the left AND right hand sides. So there needs to be some kind of cross feed at work. I've tried in v...


Fast cross correlation!

Started by Atmapuri in comp.dsp14 years ago 3 replies

Hi! I would like to speed up a cross correlation application which looks like this: x has to be cross-correlated with "n"...

Hi! I would like to speed up a cross correlation application which looks like this: x has to be cross-correlated with "n" different length y's. I would like to precompute the FFT of the x before making the convolution with all the y's. Thats why I dont want to use the classic convolution and cross correlation functions available in Matlab. (nor the filtering functions). Any sugg...


newbie question on discrete convolution

Started by mistvan in comp.dsp10 years ago 3 replies

Hi I am playing around with fir filters/convolution. I have a length 11 filter. I convolve this with my input signal and plot both input...

Hi I am playing around with fir filters/convolution. I have a length 11 filter. I convolve this with my input signal and plot both input and output. I see the output has a delay of 5 samples (filter length/2) with respect to the input Why is this?


FFT convolution with overlap add

Started by Uli in comp.dsp14 years ago 10 replies

Hi there, I need some advise for a convolution problem: I have to compute several thousands of convolutions. I managed to do it with the...

Hi there, I need some advise for a convolution problem: I have to compute several thousands of convolutions. I managed to do it with the overlap add method described in chapter 18 of the dspguide. But it takes too much time for computation. The method described is breaking a signal with a big number of points into small segments. Then it is convoluted by a small filter. In my case I...


Convolution and IFFT

Started by tomcat123 in comp.dsp11 years ago 3 replies

Hi, all, I am new to DSP. I try to verify the convolution in time domain equal to the mulitplication in freq domain. I have a input signal of...

Hi, all, I am new to DSP. I try to verify the convolution in time domain equal to the mulitplication in freq domain. I have a input signal of 128 points and a filter of 30 points. I do FFT of the input signal and filter(pad zero). end up with 128 points of both signal in freq domain. I then do multiplication of the two. In time domain, I do convolution of the input signal and the filter, which en...


Convolution cycles

Started by carlferns in comp.dsp13 years ago 1 reply

I am using TI's imaging libraries to do a 3x3 convolution. The documentation states the following "Cycles = 9 * cols/2 + 44 For cols = 256,...

I am using TI's imaging libraries to do a 3x3 convolution. The documentation states the following "Cycles = 9 * cols/2 + 44 For cols = 256, cycles = 1196". I see a much higher cycle count (factor of 100 over) when I profile even the simplest use of this function on a DSK6416. Code : begin = clock (); IMG_conv_3x3 (in_data, out_data, cols, mask, shift); end = clock (); ------ (en...


Convolution cycles

Started by carlferns in comp.dsp13 years ago

I am using TI's imaging libraries to do a 3x3 convolution. The documentation states the following "Cycles = 9 * cols/2 + 44 For cols = 256,...

I am using TI's imaging libraries to do a 3x3 convolution. The documentation states the following "Cycles = 9 * cols/2 + 44 For cols = 256, cycles = 1196". I see a much higher cycle count (factor of 100 over) when I profile even the simplest use of this function on a DSK6416. Code : begin = clock (); IMG_conv_3x3 (in_data, out_data, cols, mask, shift); end = clock (); ------ (en...


A Series of Convolution of Infinite Sequences

Started by cjlam in comp.dsp11 years ago 4 replies

I have a set of $M$ infinite data sequences, and I hope to perform a series of convolutions such that SEQ_result = SEQ_1 * SEQ_2 * SEQ_3 * ......

I have a set of $M$ infinite data sequences, and I hope to perform a series of convolutions such that SEQ_result = SEQ_1 * SEQ_2 * SEQ_3 * ... SEQ_M I am only interested at the first few terms of SEQ_result. Right now, I can only think of one method, which is to use fast convolution (truncation all to finite sequences, FFT, pointwise multiply, iFFT). Is there other efficient way to imple...


Discrete Wavelet Transform

Started by TheMadScientist in comp.dsp10 years ago 2 replies

I have heard the DWT requires only N calculations. The documents I have read say it uses FIR filters so wouldn't this mean at each stage...

I have heard the DWT requires only N calculations. The documents I have read say it uses FIR filters so wouldn't this mean at each stage N*N calculations are required? Even if the signal is decimated at each stage how can you be using FIR filters (convolution) and not end up with a large number of calculations? The only thing that would reduce the calculations would be to use FFT convolution bu...


Probability densities

Started by Anonymous in comp.dsp7 years ago 5 replies

Suppose I add two noise signals y=x+v where x has a PDF Px and v has a PDF Pv What is the PDF of y?? I am led to believe that it is the...

Suppose I add two noise signals y=x+v where x has a PDF Px and v has a PDF Pv What is the PDF of y?? I am led to believe that it is the convolution of the two PDF's of x and v - is this right and if so why? Therefore if x and v are both Guassian, y must have a PDF which is the convolution of two Guassian PDF's - I assume that must be guassian too???


Mic Modelling with Fast Convolution

Started by benkbenkbenk in comp.dsp12 years ago 9 replies

Hi, first of all hello, im new here. I have been trying to develop some microphone modelling software that uses impulse responses to apply the...

Hi, first of all hello, im new here. I have been trying to develop some microphone modelling software that uses impulse responses to apply the frequency response of a mic to a sound. I'm using fast convolution and the overlap-add method. I have developed the following code to convolve a signal with an impulse of less that 1024points. I'm presently testing using white noise as my signal. T...


Convolve exponential with rect

Started by Peter Mairhofer in comp.dsp5 years ago 2 replies

Hi, I have a pretty trivial issue: I want to convolve a shifted rect function with an exponential: h(t) = 2pi/K e^(-2pi/K t) p(t) =...

Hi, I have a pretty trivial issue: I want to convolve a shifted rect function with an exponential: h(t) = 2pi/K e^(-2pi/K t) p(t) = rect(t-1/2) g(t) = (h*p)(t) When I do the convolution with a discrete-time approximation numerically in MATLAB (either conv or cconv) I get the blue line: http://snag.gy/ZOhp5.jpg When do the convolution analytically, I get g(t) = e^(-2pi/K (t-...