Questions tagged [fast-convolution]
Fast convolution is using the FFT, multiplication in the frequency domain, and inverse FFT to perform convolution of a long signal with a long FIR.
57
questions
5
votes
0
answers
42
views
Avoid circular convolution aliasing in weighted overlap-add (WOLA) method
This link gives the procedure of implementing the WOLA method. It can be summarized briefly as windowing the extractd data frame, FFT, modify the spectrum, IFFT and overlap-add the windowed IFFT ...
- 279
9
votes
2
answers
1k
views
FFT-based fast convolution vs IIR filtering
I'm doing an multichannel audio equalizer system on a chip, which is parametric EQ, usually implemented in cascaded biquad IIR filters.
My problem is that I use a lot of IIR filters due to multiple ...
- 417
5
votes
3
answers
694
views
Why is sweeping in convolution so confusing
From this formula, I thought that time constant (m or tau) is the variable sweeping from -infinity to infinity.
But in this visualisation https://lpsa.swarthmore.edu/Convolution/CI.html, it is t the ...
0
votes
3
answers
65
views
Signature for the kernels in fast-convolution
Continuing the discussion from Convolution kernel in Bluestein's algorithm it has a very specific signature of the kernel
...
- 571
0
votes
0
answers
36
views
circular convolution using fftw: How to make the sequence periodic?
I have a grid size of $(4,4,4)$ as shown the picture below.
$G_m$
Each box/group contains $\ell$ samples and each sample is a complex $\mathrm{3d-Vector}$
Now I define $G_m$ for each box/group, ...
- 571
0
votes
1
answer
31
views
How to zero pad a 3d data and apply fast convolution?
I have two sequences,
Let $A$ be the first sequence whose dimensions are $(5, 5, 3)$ takes
complex values
Let $B$ be the second sequence whose dimensions are $(5, 5, 1)$ also takes complex values
I ...
- 571
0
votes
1
answer
46
views
What means `crop` in FFT calculation?
In soapy power manual:
Crop: -o PERCENT, --overlap PERCENT
percent of overlap when frequency hopping (incompatible with -k)
-k PERCENT, --crop PERCENT
percent of crop when frequency hopping (...
- 175
1
vote
3
answers
534
views
Are there any reasons to use overlap algorithm when one could do ifft(X * Y) from the complete signal as well in similiar time complexity
We learned that overlap add and overlap save are used as fast convolution methods, because they can be applied with fft due to the formula $x[n] * y[n] = \text{IFFT}(X[k] Y[k])$
So now I was wondering ...
- 209
1
vote
1
answer
114
views
Convolve audio signal with filter with different sampling rates
I would like to convolve two signals:
Room impulse response that was obtained by deconvolution, measured at 48 kHz. Truncated to 8192 samples.
Anechoic recording (.wav) recorded at 41.1 kHz
If I ...
- 13
0
votes
1
answer
58
views
Result of convolution using FFT method not accurate in time
I am trying to implement fast convolution between a signal and complex Morlet wavelets. To make the result equivalent to a linear convolution, I let the MATLAB fft function zero-pad both the signal ...
5
votes
1
answer
139
views
0
votes
0
answers
48
views
Subband Tuning in Matlab
Suppose you have an i/q stream with a sample rate of 20 MHz. You want to bandpass filter a 1 MHz bandwidth (ie passband from 5 to 6 MHz). You then want to heterodyne that 1 MHz chunk. How would you do ...
- 956
-1
votes
1
answer
91
views
FFT for Arbitrary time history
I have two Time-Speed histories one is real ( real measurements) the other is synthetically generated using formula. I need to compare both in the frequency domain using FFT given that:
the time step ...
- 1
6
votes
2
answers
279
views
Explain the Process of Spectral Pooling and Spectral Activation in the Context of CNN in Frequency Domain
I am reading the paper Design of an energy efficient accelerator for training of convolutional neural networks using frequency Domain Computation:
which uses Frequency Pooling, from Spectral ...
- 333
2
votes
0
answers
87
views
Multiple 1D gaussian filter
Given
$$
\begin{cases}
&f_0(x)=1 \\
&f_{n+1}(x)= (\varphi*(f\mathbb{I}_{[a_n,b_n]}))(x)=\int_{-\infty}^{+\infty}\varphi(x-t)f_n(t)\mathbb{I}_{[a_n,b_n]}(t)dt = \int_{a_n}^{b_n}\varphi(x-t)f_n(...
- 143
2
votes
1
answer
309
views
Fast computation of a convolution integral with Gaussian kernel
Given a convolution integral with gaussian kernel
$$
g(y) =\int_a^b\varphi(y-x)f(x)dx=\int_{-\infty}^{+\infty}\varphi(y-x)f(x)\mathbb{I}_{[a,b]}(x)dx
$$
where
$\varphi(x)= \frac{1}{\sqrt{2\pi}}\exp{\...
- 143
2
votes
1
answer
189
views
ringing artifacts using FFT-based gaussian blurring
I'm trying to do an FFT-based gaussian blur on a grayscale image, and it works, however it seems to introduce ringing artifacts to the result when compared to the expected direct filter. What can I do ...
- 23
-1
votes
3
answers
149
views
Role of windowing in implementing FIR filters using Fast Convolution
In Fast Convolution the filtering is performed by taking FFT of both the signal and impulse response and multiplication in frequency domain using Overlap add/Overlap save method for processing ...
- 386
4
votes
2
answers
164
views
Why is FFT-based convolution efficient only for signals of large size?
According to the documentation of scipy.signal.fftconvolve
This is generally much faster than convolve for large arrays (n >
~500), but can be slower when only ...
- 85
5
votes
3
answers
2k
views
Can FFT convolution be faster than direct convolution for signals of large sizes?
Let's say I have a 1D signal of size $N$ and am trying to filter it with a 10-tap FIR filter mask. When $N$ is large, the number of multiply-accumulates would approximately equal
$$2 \times 10 \times ...
- 757
1
vote
1
answer
319
views
Computing Autocorrelation via FFT
I have seen several methods to calculate Autocorrelations using FFTs, and am confused about why they differ.
Zero-Pad it to double its original length.Take the FFT. Then replace all the coefficients ...
- 338
0
votes
1
answer
66
views
Confusion regarding plot of linear convolution vs fast convolution via FFT
Question
[Example 5.23 of Digital Signal Processing Using MATLAB Third Edition By Vinay K. Ingle and John G. Proakis ]
I am not getting same plot/results as book ...
- 931
6
votes
0
answers
176
views
Efficient Convolution Without I/O Delay (Gardner 1995) implementation
I am trying to perform real-time convolution on an audio stream using a fairly large FIR for convolutional reverb.
I found a great paper explaining how to do just that: Gardner, 1995. But, I'm new to ...
- 161
0
votes
0
answers
360
views
Handling stride>1 in FFT-based convolution
The Fourier transform of the convolution of two signals with stride 1 is equivalent to point-wise multiplication of their individual Fourier transforms. I need to perform stride-'n' convolution using ...
- 71
1
vote
1
answer
291
views
How do I pick an optimal window size for overlap add/save method algorithmically
Currently I've implemented an overlap-add method using resources on the internet, but I couldn't find a well documented way to minimize the cost of the method. In other words, how do I pick the window ...
- 111
1
vote
1
answer
1k
views
How to manually implement convolution with FFTs?
I'm trying to manually implement a convolution using FFTs and it isn't working as expected. I know I'm probably missing some subtlety with padding, shifting, or conjugation, (all of which I've tried ...
- 201
3
votes
0
answers
78
views
Understanding convolution of Chirp z algorithm
I dont´t understand how works the convolution part of the Chirp z.
I understand how the DFT is transformed
\begin{align*}
x(k) = \sum_{n=0}^{N-1} x(n) W_N^{kn}
\end{align*}
to this expresion:
\...
- 31
1
vote
1
answer
603
views
How does minimum-latency partitioned convolution reverb work when you receive input samples in chunks, rather than one at a time?
I'm writing a reverb system where I receive an input block of samples 480 elements long, do some operation on them, and pass the block on to the next effect.
I've been reading up on partitioned ...
- 27
2
votes
1
answer
679
views
Overlap Add - which length to use?
I am doing realtime audio. Let's say I have a signal $x(n)$ which is passed in blocks of 512 (siobuffer) to the program filter/FFT method. Within this block I want to convolve the signal with the ...
- 51
2
votes
0
answers
134
views
Correct way of derivating in frequency domain with FFT
I believe I am very close to the answer and only need a small nudge to get to the answer.
What I want:
I want to take a signal, use FFT to transform it to the frequency domain (FD), multiply it by $...
- 216
6
votes
1
answer
460
views
What is the Fastest Way Ever to Do 2D Correlation?
Here is what i gathered so far:
Any sliding window classification, image filtering or similar can be fastly done by a FFT (flip the signal and do convolution).
Also, if the template/filter kernel is ...
3
votes
0
answers
84
views
Convolution of signals sampled on a logarithmic grid
Is there a practical accelerated algorithm or a theoretical discrete (Fourier) transform based method to convolve discrete-time signals sampled on a logarithmic grid? What I mean is representing a ...
- 12.5k
1
vote
1
answer
188
views
Perform overlap add with available FFT size smaller than filter coefficients
I would like to process 12 to 20 seconds of incoming audio at a sample rate of 44100. I must process this audio in real time in an STM embedded kit (perhaps also an Android Smartphone). I'm trying to ...
- 962
1
vote
2
answers
904
views
Linear Convolution in DSP and Hann window
My understanding is that when dealing with discrete finite sequences, convolution done thru FFT is circular. To obtain linear convolution, one would pad the input with zeros up some appropriate length....
0
votes
2
answers
99
views
How to detect a long signal inside an even longer one?
I'm trying to detect a long pattern inside a long recorded audio. In order to do so I am implementing a matched filter in time by performing a FIR filtering using the coefficients of my pattern ...
- 962
1
vote
1
answer
2k
views
Decimation and filtering in the frequency domain
I'm working on a Software Defined Radio project where I'd like to low-pass filter and decimate an analytical signal (IQ) sampled at 96ksps. Let's say the low-pass filter has a cutoff at 5kHz and I'd ...
- 187
1
vote
1
answer
74
views
position of useful filtered data inside DFT based filtered output
I'm trying to filter a 400 samples signal with various bandstop FIR filters (constant group delay), one at a time to see which one gives me the desired result. Every filter was build using the Kaiser ...
- 962
9
votes
5
answers
2k
views
Fast & accurate convolution algorithm (like FFT) for high dynamic range?
It seems that FFT-based convolution suffers from limited floating-point resolution due to evaluating everything around the roots of unity, as you can see in the $10^{14}$-factor error in this Python ...
- 473
0
votes
1
answer
367
views
Amplitude not attenuated after bandpass filtering?
I want to make a bandpass filter for an IQ signal recorded from an Software Defined Radio with center frequency "Fc" & sampling frequency "Fs". For futhur analysis I wanted to focus upon signal ...
- 47
1
vote
0
answers
41
views
Can this robot trajectory smoothing function be expressed as a single convolution?
I have an Nx3 matrix of poses (x, y, theta) that I need to apply an algorithm on which will cause the trajectory to be smoothed. I am interested in performance and so would like to improve efficiency ...
- 107
0
votes
2
answers
471
views
Can I compute convolution through the overlap and add method without the IFFT?
So, I'm (I think) aware of how the overlap and add algorithm for linear convolution works, but my question is that, suppose I have a FFT-ed set of sequences that belong to a large sequence.
Can I add ...
3
votes
2
answers
407
views
Understanding DFT-ODD operation in Gardner Efficient Convolution paper
As a newbie to DSP I am trying to understand and implement an efficient convolution engine as per JAES V43.3 1995/03 - Gardner - Efficient Convolution without I/O Delay.
I'm pretty much there with a ...
- 215
2
votes
3
answers
681
views
Could there be any reason to prefer convolution-based calculation of autocorrelation?
Theoretically both of ways of calculating autocorrelation function are identical: strightforward convolution and Fourier-based method where we use FFT/iFFT in practice. And as it is well known, the ...
- 123
6
votes
1
answer
4k
views
Implementing Convolution in Frequency Domain?
Suppose, we have a bitmap image represented as a 2D integer array,
int [,] image2D; whose FFT is Complex[,] fftImage2D;
...
user18425
0
votes
1
answer
290
views
Slight crackle with overlap-add frequency-domain convolution
I'm trying to make an FFT audio filter using overlap-add but there's some slight rhythmic distortion/crackle in the output. I have an irSize (impulse response size) of 512 samples and a blockSize of ...
1
vote
0
answers
765
views
Overlap add / overlap save using FFTW
I'm working on a realtime FFT processing algorithm in C using PortAudio. Currently I'm just trying to test if I can do OLA / OLS in (semi) real time using FFTW. I'm getting some distortion in the ...
- 11
2
votes
2
answers
311
views
Fast convolution with very high order FIR
I am investigating the overlap-add and overlap-save methods for processing an audio signal with a FIR. The FIR is a measured impulse response of a reverberant space and may be of order greater than ...
3
votes
2
answers
1k
views
Frequency Domain Filtering with big kernel size
I use FFT to do filtering in frequency domain, but when I use big kernel I got shift on border, I think it's due to nature of cyclical convolution and I need to do more zero padding.
But how to ...
- 550
3
votes
2
answers
2k
views
direct Frequency domain FIR filtering vs Overlap-add method
I'm trying to do bandpass filtering of a EEG signal samples at 250Hz and benchmarking the following 4 methods of FIR filtering for different filter orders. The length of the signal is 15000 samples.
...
- 418
0
votes
2
answers
390
views
Running an FFT filter on a large data set
I have a non realtime application where I need to run a bandpass FFT filter on a data array of between 5k and 10k data points. Do I break it up into (say) 256 point chunks, run the FFT on that and ...
- 177