Related
The Problem
On a server, I host ids in a json file. From clients, I need to mandate the server to intersect and sometimes negate these ids (the ids never travel to the client even though the client instructs the server its operations to perform).
I typically have 1000's of ids, often have 100,000's of ids, and have a maximum of 56,000,000 of them, where each value is unique and between -100,000,000 and +100,000,000.
These ids files are stable and do not change (so it is possible to generate a different representation for it that is better adapted for the calculations if needed).
Sample ids
Largest file sizes
I need an algorithm that will intersect ids in the sub-second range for most cases. What would you suggest? I code in java, but do not limit myself to java for the resolution of this problem (I could use JNI to bridge to native language).
Potential solutions to consider
Although you could not limit yourselves to the following list of broad considerations for solutions, here is a list of what I internally debated to resolve the situation.
Neural-Network pre-qualifier: Train a neural-network for each ids list that accepts another list of ids to score its intersection potential (0 means definitely no intersection, 1 means definitely there is an intersection). Since neural networks are good and efficient at pattern recognition, I am thinking of pre-qualifying a more time-consuming algorithm behind it.
Assembly-language: On a Linux server, code an assembly module that does such algorithm. I know that assembly is a mess to maintain and code, but sometimes one need the speed of an highly optimized algorithm without the overhead of a higher-level compiler. Maybe this use-case is simple enough to benefit from an assembly language routine to be executed directly on the Linux server (and then I'd always pay attention to stick with the same processor to avoid having to re-write this too often)? Or, alternately, maybe C would be close enough to assembly to produce clean and optimized assembly code without the overhead to maintain assembly code.
Images and GPU: GPU and image processing could be used and instead of comparing ids, I could BITAND images. That is, I create a B&W image of each ids list. Since each id have unique values between -100,000,000 and +100,000,000 (where a maximum of 56,000,000 of them are used), the image would be mostly black, but the pixel would become white if the corresponding id is set. Then, instead of keeping the list of ids, I'd keep the images, and do a BITAND operation on both images to intersect them. This may be fast indeed, but then to translate the resulting image back to ids may be the bottleneck. Also, each image could be significantly large (maybe too large for this to be a viable solution). An estimate of a 200,000,000 bits sequence is 23MB each, just loading this in memory is quite demanding.
String-matching algorithms: String comparisons have many adapted algorithms that are typically extremely efficient at their task. Create a binary file for each ids set. Each id would be 4 bytes long. The corresponding binary file would have each and every id sequenced as their 4 bytes equivalent into it. The algorithm could then be to process the smallest file to match each 4 bytes sequence as a string into the other file.
Am I missing anything? Any other potential solution? Could any of these approaches be worth diving into them?
I did not yet try anything as I want to secure a strategy before I invest what I believe will be a significant amount of time into this.
EDIT #1:
Could the solution be a map of hashes for each sector in the list? If the information is structured in such a way that each id resides within its corresponding hash key, then, the smaller of the ids set could be sequentially ran and matching the id into the larger ids set first would require hashing the value to match, and then sequentially matching of the corresponding ids into that key match?
This should make the algorithm an O(n) time based one, and since I'd pick the smallest ids set to be the sequentially ran one, n is small. Does that make sense? Is that the solution?
Something like this (where the H entry is the hash):
{
"H780" : [ 45902780, 46062780, -42912780, -19812780, 25323780, 40572780, -30131780, 60266780, -26203780, 46152780, 67216780, 71666780, -67146780, 46162780, 67226780, 67781780, -47021780, 46122780, 19973780, 22113780, 67876780, 42692780, -18473780, 30993780, 67711780, 67791780, -44036780, -45904780, -42142780, 18703780, 60276780, 46182780, 63600780, 63680780, -70486780, -68290780, -18493780, -68210780, 67731780, 46092780, 63450780, 30074780, 24772780, -26483780, 68371780, -18483780, 18723780, -29834780, 46202780, 67821780, 29594780, 46082780, 44632780, -68406780, -68310780, -44056780, 67751780, 45912780, 40842780, 44642780, 18743780, -68220780, -44066780, 46142780, -26193780, 67681780, 46222780, 67761780 ],
"H782" : [ 27343782, 67456782, 18693782, 43322782, -37832782, 46152782, 19113782, -68411782, 18763782, 67466782, -68400782, -68320782, 34031782, 45056782, -26713782, -61776782, 67791782, 44176782, -44096782, 34041782, -39324782, -21873782, 67961782, 18703782, 44186782, -31143782, 67721782, -68340782, 36103782, 19143782, 19223782, 31711782, 66350782, 43362782, 18733782, -29233782, 67811782, -44076782, -19623782, -68290782, 31721782, 19233782, 65726782, 27313782, 43352782, -68280782, 67346782, -44086782, 67741782, -19203782, -19363782, 29583782, 67911782, 67751782, 26663782, -67910782, 19213782, 45992782, -17201782, 43372782, -19992782, -44066782, 46142782, 29993782 ],
"H540" : [...
You can convert each file (list of ids) into a bit-array of length 200_000_001, where bit at index j is set if the list contains value j-100_000_000. It is possible, because the range of id values is fixed and small.
Then you can simply use bitwise and and not operations to intersect and negate lists of ids. Depending on the language and libraries used, it would require operating element-wise: iterating over arrays and applying corresponding operations to each index.
Finally, you should measure your performance and decide whether you need to do some optimizations, such as parallelizing operations (you can work on different parts of arrays on different processors), preloading some of arrays (or all of them) into memory, using GPU, etc.
First, the bitmap approach will produce the required performance, at a huge overhead in memory. You'll need to benchmark it, but I'd expect times of maybe 0.2 seconds, with that almost entirely dominated by the cost of loading data from disk, and then reading the result.
However there is another approach that is worth considering. It will use less memory most of the time. For most of the files that you state, it will perform well.
First let's use Cap'n Proto for a file format. The type can be something like this:
struct Ids {
is_negated #0 :Bool;
ids #1 :List(Int32);
}
The key is that ids are always kept sorted. So list operations are a question of running through them in parallel. And now:
Applying not is just flipping is_negated.
If neither is negated, it is a question of finding IDs in both lists.
If the first is not negated and the second is, you just want to find IDs in the first that are not in the second.
If the first is negated and the second is not, you just want to find IDs in the second that are not in the first.
If both are negated, you just want to find all ids in either list.
If your list has 100k entries, then the file will be about 400k. A not requires copying 400k of data (very fast). And intersecting with another list of the same size involves 200k comparisons. Integer comparisons complete in a clock cycle, and branch mispredictions take something like 10-20 clock cycles. So you should be able to do this operation in the 0-2 millisecond range.
Your worst case 56,000,000 file will take over 200 MB and intersecting 2 of them can take around 200 million operations. This is in the 0-2 second range.
For the 56 million file and a 10k file, your time is almost all spent on numbers in the 56 million file and not in the 10k one. You can speed that up by adding a "galloping" mode where you do a binary search forward in the larger file looking for the next matching number and picking most of them. Do be warned that this code tends to be tricky and involves lots of mispredictions. You'll have to benchmark it to find out how big a size difference is needed.
In general this approach will lose for your very biggest files. But it will be a huge win for most of the sizes of file that you've talked about.
I am doing some interesting experiments with audio and image files and Fast-Fourier Transforms (FFTs).
Fast Fourier Transforms are used in signal processing rather than other Fourier Transform algorithms because for large quantities of data they are the only (or one of the only) viable algorithm variants to use, as they scale as O(n log(n)), rather than n^2 as the naive implementation does.
The disadvantage is that the data must be stored in an array which has 2^n elements, for n integer.
When processing some data which does not have 2^n elements, the simple approach is to extend the array to be length 2^n and fill the "empty" elements with zero. (Assuming the mean value of the input signal is zero.)
I wrote a program to process some audio samples taken from WAV files. I tried implementing things such as a low-cut filter. In this case I found that my output signal (after doing the reverse transform) cuts to zero amplitude after a certain period of time. This is obviously not what one would expect of a low-pass filter.
I could dump my code at this point, but that is neither useful, nor legal as the source of my algorithm is a text-book with closed source code.
Instead I shall ask the following question.
Is packing out the array with zeros the best possible thing to do? Could this be causing my program to produce the unexpected results I am seeing? if I understand fourier mathematics correctly, having a bunch of zeros at the end of my array will introduce a large amount of low and high-frequency content as this essentially looks like a step-function (low frequency square wave). Should I be doing something else such as implementing my band-pass filter in a different way, for example, splitting the data into smaller groups of say 1024 samples and applying the FT, filter and IFT (inverse FT) to those small groups?
This question has been tagged with theory as it is not related to any specific programming language. (I assume that is the correct tag to use?)
Edit: It's now working beautifully, thanks all, I was able to pinpoint the 2 mistakes I made using the information below.
All finite length DFTs and FFT multiply longer data (longer source data or wav file than the FFT) with a rectangular window, which convolves the spectrum with a (periodic) Sinc function. Zero padding uses a shorter rectangular window, which results in the convolution of the spectrum with a wider Sinc function.
Filtering by multiplication of FFTs results in circular convolution, which wraps the impulse response of the filter around the FFT/IFFT result (e.g. the end of your filtered signal will interfere with the beginning of the filtered signal within the IFFT result). So you want to zero-pad your data before the FFT, and then see the impulse response of your filter go to zero at or before the very end of the filtered result (e.g. not wrap around). Look up the overlap-add and overlap-save algorithms, for using short FFTs for fast convolution filtering of longer signals, which take care of the filter impulse response extending into the zero-padded portion.
You can also use FFTs that are not a power of 2 in length. Any length that can be factored into small primes will work with most modern FFT libraries.
It depends what you are interested in.
If you are just interested in spectrum magnitude, then place the real data in the middle of the window to be processed. Just know that this time shift will put a phase shift into the spectrum result.
Regardless of the number of points, do not forget to place a window on your data. Wikipedia has a good write up on the windowing functions at https://en.wikipedia.org/wiki/Window_function.
If you do not perform some sort of windowing on your real world data, the padded signal will appear to have a step up and a step down at the end of the valid data (which puts a lot of noise into your spectrum giving you the false impression that you have a noise floor).
So, my recommendation, if you primarily care about magnitude:
- develop a hamming window for the number of points of valid data you have.
- apply the hamming window to the data you have
After that you have OPTIONS:
A) if your samples are slightly above a base two number, use the lower base two number (i.e. if you have 1400 points, do two 1024 point FFTs with overlap). The results of these two FFTs can be "smartly" combined for an aggregate spectrum. Depending on your fidelity needs, you can do this with more FFTs with a larger portion of overlapped data. Try to keep the overlap less that 10% to account for your window edges that will get attenuated by the start and end of the windowing functions.
B) place your windowed data anywhere in the FFT input vector (beginning, middle or end, it should only impact your phase results - which is why I asked if phase is important).
If it turns out phase is important, start your valid windowed data at the beginning of the FFT vector.
Regarding your spectrum observations (I just went through the same thing two weeks ago). If you are looking at a wave file converted from a lossy compression, you are going to be starting with a band limited signal, so expect the spectrum to do an abrupt drop. My first lossless wave file plot had a huge bald spot from Fs/10 -> 9Fs/10 (which is expected). For your plots - also display your data in logarithmic bins (linear bins will give you misleading info and squish the lower frequency elements which are the bulk of the signal in compressed music files).
FYI - I recommended hamming (because I did the same thing). A decoded compressed audio signal will only use a portion of your spectrum (decoding a 320kbps stream is sampled at 10Khz), even when decoded to 44.1Khz representation, all of the interesting data should be below 5Khz.
Best of luck
J.R.
P.S. this is my first post here, chime back if you want some pretty pictures from TeraPlot.
This is a question for http://dsp.stackexchange.com but yes, zero-padding is perfectly legitimate here.
Here’s why the filtered signal (once it’s back in the time-domain) goes to zero after some time: imagine linearly-convolving the zero-padded signal with your low-pass filter’s impulse response (using the slow O(N^2) time-domain filter implementation). The output will go to zero after the original signal is done, when the filter is just being fed with zeros, right? That result will be the same as the output of FFT-based fast convolution. It’s perfectly normal. Just crop the output signal to the same length of the input and move on with your life.
Caveat on FFT orders: just because power-of-two FFT lengths are “the fastest” in terms of operation count, while FFTs of lengths with low prime factors (3, 5, 7) have slightly higher operation counts, you may find that zero-padding to a low-prime-factor is faster in terms of real-world runtime because of memory costs. A pathological example: if you have a 1025-long signal, you probably don’t want to zero-pad to 2048 and eat the cost of allocating a nearly 2x memory buffer, and running a nearly 2x longer FFT. You’d try 1080-length FFT or something (1080 = 2^3 * 3^3 * 5: nextprod is your friend) and wouldn’t be surprised if it completed much faster than power-of-two.
I have a long time series with some repeating and similar looking signals in it (not entirely periodical). The length of the time series is about 60000 samples. To identify the signals, I take out one of them, having a length of around 1000 samples and move it along my timeseries data sample by sample, and compute cross-correlation coefficient (in Matlab: corrcoef). If this value is above some threshold, then there is a match.
But this is excruciatingly slow (using 'for loop' to move the window).
Is there a way to speed this up, or maybe there is already some mechanism in Matlab for this ?
Many thanks
Edited: added information, regarding using 'xcorr' instead:
If I use 'xcorr', or at least the way I have used it, I get the wrong picture. Looking at the data (first plot), there are two types of repeating signals. One marked by red rectangles, whereas the other and having much larger amplitudes (this is coherent noise) is marked by a black rectangle. I am interested in the first type. Second plot shows the signal I am looking for, blown up.
If I use 'xcorr', I get the third plot. As you see, 'xcorr' gives me the wrong signal (there is in fact high cross correlation between my signal and coherent noise).
But using "'corrcoef' and moving the window, I get the last plot which is the correct one.
There maybe a problem of normalization when using 'xcorr', but I don't know.
I can think of two ways to speed things up.
1) make your template 1024 elements long. Suddenly, correlation can be done using FFT, which is significantly faster than DFT or element-by-element multiplication for every position.
2) Ask yourself what it is about your template shape that you really care about. Do you really need the very high frequencies, or are you really after lower frequencies? If you could re-sample your template and signal so it no longer contains any frequencies you don't care about, it will make the processing very significantly faster. Steps to take would include
determine the highest frequency you care about
filter your data so higher frequencies are blocked
resample the resulting data at a lower sampling frequency
Now combine that with a template whose size is a power of 2
You might find this link interesting reading.
Let us know if any of the above helps!
Your problem seems like a textbook example of cross-correlation. Therefore, there's no good reason using any solution other than xcorr. A few technical comments:
xcorr assumes that the mean was removed from the two cross-correlated signals. Furthermore, by default it does not scale the signals' standard deviations. Both of these issues can be solved by z-scoring your two signals: c=xcorr(zscore(longSig,1),zscore(shortSig,1)); c=c/n; where n is the length of the shorter signal should produce results equivalent with your sliding window method.
xcorr's output is ordered according to lags, which can obtained as in a second output argument ([c,lags]=xcorr(..). Always plot xcorr results by plot(lags,c). I recommend trying a synthetic signal to verify that you understand how to interpret this chart.
xcorr's implementation already uses Discere Fourier Transform, so unless you have unusual conditions it will be a waste of time to code a frequency-domain cross-correlation again.
Finally, a comment about terminology: Correlating corresponding time points between two signals is plain correlation. That's what corrcoef does (it name stands for correlation coefficient, no 'cross-correlation' there). Cross-correlation is the result of shifting one of the signals and calculating the correlation coefficient for each lag.
iOS has an issue recording through some USB audio devices. It cannot be reliably reproduced (happens every 1 in ~2000-3000 records in batches and silently disappears), and we currently manually check our audio for any recording issues. It results in small numbers of samples (1-20) being shifted by a small number that sounds like a sort of 'crackle'.
They look like this:
closer:
closer:
another, single sample error elsewhere in the same audio file:
The question is, how can these be algorithmically be detected (assuming direct access to samples) whilst not triggering false positives on high frequency audio with waveforms like this:
Bonus points: after determining as many errors as possible, how can the audio be 'fixed'?
Dirty audio file - pictured
Another dirty audio file
Clean audio with valid high frequency - pictured
More bonus points: what could be causing this issue in the iOS USB audio drivers/hardware (assuming it is there).
I do not think there is an out of the box solution to find the disturbances, but here is one (non standard) way of tackling the problem. Using this, I could find most intervals and I only got a small number of false positives, but the algorithm could certainly use some fine tuning.
My idea is to find the start and end point of the deviating samples. The first step should be to make these points stand out more clearly. This can be done by taking the logarithm of the data and taking the differences between consecutive values.
In MATLAB I load the data (in this example I use dirty-sample-other.wav)
y1 = wavread('dirty-sample-pictured.wav');
y2 = wavread('dirty-sample-other.wav');
y3 = wavread('clean-highfreq.wav');
data = y2;
and use the following code:
logdata = log(1+data);
difflogdata = diff(logdata);
So instead of this plot of the original data:
we get:
where the intervals we are looking for stand out as a positive and negative spike. For example zooming in on the largest positive value in the plot of logarithm differences we get the following two figures. One for the original data:
and one for the difference of logarithms:
This plot could help with finding the areas manually but ideally we want to find them using an algorithm. The way I did this was to take a moving window of size 6, computing the mean value of the window (of all points except the minimum value), and compare this to the maximum value. If the maximum point is the only point that is above the mean value and at least twice as large as the mean it is counted as a positive extreme value.
I then used a threshold of counts, at least half of the windows moving over the value should detect it as an extreme value in order for it to be accepted.
Multiplying all points with (-1) this algorithm is then run again to detect the minimum values.
Marking the positive extremes with "o" and negative extremes with "*" we get the following two plots. One for the differences of logarithms:
and one for the original data:
Zooming in on the left part of the figure showing the logarithmic differences we can see that most extreme values are found:
It seems like most intervals are found and there are only a small number of false positives. For example running the algorithm on 'clean-highfreq.wav' I only find one positive and one negative extreme value.
Single values that are falsely classified as extreme values could perhaps be weeded out by matching start and end-points. And if you want to replace the lost data you could use some kind of interpolation using the surrounding data-points, perhaps even a linear interpolation will be good enough.
Here is the MATLAB-code I used:
function test20()
clc
clear all
y1 = wavread('dirty-sample-pictured.wav');
y2 = wavread('dirty-sample-other.wav');
y3 = wavread('clean-highfreq.wav');
data = y2;
logdata = log(1+data);
difflogdata = diff(logdata);
figure,plot(data),hold on,plot(data,'.')
figure,plot(difflogdata),hold on,plot(difflogdata,'.')
figure,plot(data),hold on,plot(data,'.'),xlim([68000,68200])
figure,plot(difflogdata),hold on,plot(difflogdata,'.'),xlim([68000,68200])
k = 6;
myData = difflogdata;
myPoints = findPoints(myData,k);
myData2 = -difflogdata;
myPoints2 = findPoints(myData2,k);
figure
plotterFunction(difflogdata,myPoints>=k,'or')
hold on
plotterFunction(difflogdata,myPoints2>=k,'*r')
figure
plotterFunction(data,myPoints>=k,'or')
hold on
plotterFunction(data,myPoints2>=k,'*r')
end
function myPoints = findPoints(myData,k)
iterationVector = k+1:length(myData);
myPoints = zeros(size(myData));
for i = iterationVector
subVector = myData(i-k:i);
meanSubVector = mean(subVector(subVector>min(subVector)));
[maxSubVector, maxIndex] = max(subVector);
if (sum(subVector>meanSubVector) == 1 && maxSubVector>2*meanSubVector)
myPoints(i-k-1+maxIndex) = myPoints(i-k-1+maxIndex) +1;
end
end
end
function plotterFunction(allPoints,extremeIndices,markerType)
extremePoints = NaN(size(allPoints));
extremePoints(extremeIndices) = allPoints(extremeIndices);
plot(extremePoints,markerType,'MarkerSize',15),
hold on
plot(allPoints,'.')
plot(allPoints)
end
Edit - comments on recovering the original data
Here is a slightly zoomed out view of figure three above: (the disturbance is between 6.8 and 6.82)
When I examine the values, your theory about the data being mirrored to negative values does not seem to fit the pattern exactly. But in any case, my thought about just removing the differences is certainly not correct. Since the surrounding points do not seem to be altered by the disturbance, I would probably go back to the original idea of not trusting the points within the affected region and instead using some sort of interpolation using the surrounding data. It seems like a simple linear interpolation would be a quite good approximation in most cases.
To answer the question of why it happens -
A USB audio device and host are not clock synchronous - that is to say that the host cannot accurately recover the relationship between the host's local clock and the word-clock of the ADC/DAC on the audio interface. Various techniques do exist for clock-recovery with various degrees of effectiveness. To add to the problem, the bus clock is likely to be unrelated to either of the two audio clocks.
Whilst you might imagine this not to be too much of a concern for audio receive - audio capture callbacks could happen when there is data - audio interfaces are usually bi-directional and the host will be rendering audio at regular interval, which the other end is potentially consuming at a slightly different rate.
In-between are several sets of buffers, which can over- or under-run, which is what looks to be happening here; the interval between it happening certainly seems about right.
You might find that changing USB audio device to one built around a different chip-set (or, simply a different local oscillator) helps.
As an aside both IEEE1394 audio and MPEG transport streams have the same clock recovery requirement. Both of them solve the problem with by embedding a local clock reference packet into the serial bitstream in a very predictable way which allows accurate clock recovery on the other end.
I think the following algorithm can be applied to samples in order to determine a potential false positive:
First, scan for high amount of high frequency, either via FFT'ing the sound block by block (256 values maybe), or by counting the consecutive samples above and below zero. The latter should keep track of maximum consecutive above zero, maximum consecutive below zero, the amount of small transitions around zero and the current volume of the block (0..1 as Audacity displays it). Then, if the maximum consecutive is below 5 (sampling at 44100, and zeroes be consecutive, while outstsanding samples are single, 5 responds to 4410Hz frequency, which is pretty high), or the sum of small transitions' lengths is above a certain value depending on maximum consecutive (I believe the first approximation would be 3*5*block size/distance between two maximums, which roughly equates to period of the loudest FFT frequency. Also it should be measured both above and below threshold, as we can end up with an erroneous peak, which will likely be detected by difference between main tempo measured on below-zero or above-zero maximums, also by std-dev of peaks. If high frequency is dominant, this block is eligible only for zero-value testing, and a special means to repair the data will be needed. If high frequency is significant, that is, there is a dominant low frequency detected, we can search for peaks bigger than 3.0*high frequency volume, as well as abnormal zeroes in this block.
Also, your gaps seem to be either highly extending or plain zero, with high extends to be single errors, and zero errors range from 1-20. So, if there is a zero range with values under 0.02 absolute value, which is directly surrounded by values of 0.15 (a variable to be finetuned) or higher absolute value AND of the same sign, count this point as an error. Single values that stand out can be detected if you calculate 2.0*(current sample)-(previous sample)-(next sample) and if it's above a certain threshold (0.1+high frequency volume, or 3.0*high frequency volume, whichever is bigger), count this as an error and average.
What to do with zero gaps found - we can copy values from 1 period backwards and 1 period forwards (averaging), where "period" is of the most significant frequency of the FFT of the block. If the "period" is smaller than the gap (say we've detected a gap of zeroes in a high-pitched part of the sound), use two or more periods, so the source data will all be valid (in this case, no averaging can be done, as it's possible that the signal 2 periods forward from the gap and 2 periods back will be in counterphase). If there are more than one frequency of about equal amplitude, we can plain sample these with correct phases, cutting the rest of less significant frequencies altogether.
The outstanding sample should IMO just be averaged by 2-4 surrounding samples, as there seems to be only a single sample ever encountered in your sound files.
The discrete wavelet transform (DWT) may be the solution to your problem.
A FFT calculation is not very useful in your case since its an average representation of relative frequency content over the entire duration of the signal, and thus impossible to detect momentary changes. The dicrete short time frequency transform (STFT) tries to tackle this by computing the DFT for short consecutive time-blocks of the signal, the length of which is determine by the length (and shape) of a window, but since the resolution of the DFT is dependent on the data/block-length, there is a trade-off between resolution in freqency OR in time, and finding this magical fixed window-size can be tricky!
What you want is a time-frequency analysis method with good time resolution for high-frequency events, and good frequency resolution for low-frequency events... Enter the discrete wavelet transform!
There are numerous wavelet transforms for different applications and as you might expect, it's computationally heavy. The DWT may not be practical solution to your problem, but it's worth considering. Good luck with your problem. Some friday-evening reading:
http://klapetek.cz/wdwt.html
http://etd.lib.fsu.edu/theses/available/etd-11242003-185039/unrestricted/09_ds_chapter2.pdf
http://en.wikipedia.org/wiki/Wavelet_transform
http://en.wikipedia.org/wiki/Discrete_wavelet_transform
You can try the following super-simple approach (maybe it's enough):
Take each point in your wave-form and subtract its predecessor (look at the changes from one point to the next).
Look at the distribution of these changes and find their standard deviation.
If any given difference is beyond X times this standard deviation (either above or below), flag it as a problem.
Determine the best value for X by playing with it and seeing how well it performs.
Most "problems" should come as a pair of two differences beyond your cutoff, one going up, and one going back down.
To stick with the super-simple approach, you can then fix the data by just interpolating linearly between the last good point before your problem-section and the first good point after. (Make sure you don't just delete the points as this will influence (raise) the pitch of your audio.)
Background: I have video clips and audio tracks that I want to sync with said videos.
From the video clips, I'll extract a reference audio track.
I also have another track that I want to synchronize with the reference track. The desync comes from editing, which altered the intervals for each cutscene.
I need to manipulate the target track to look like (sound like, in this case) the ref track. This amounts to adding or removing silence at the correct locations. This could be done manually, but it'd be extremely tedious. So I want to be able to determine these locations programatically.
Example:
0 1 2
012345678901234567890123
ref: --part1------part2------
syn: -----part1----part2-----
# (let `-` denote silence)
Output:
[(2,6), (5,9) # part1
(13, 17), (14, 18)] # part2
My idea is, starting from the beginning:
Fingerprint 2 large chunks* of audio and see if they match:
If yes: move on to the next chunk
If not:
Go down both tracks looking for the first non-silent portion of each
Offset the target to match the original
Go back to the beginning of the loop
# * chunk size determined by heuristics and modifiable
The main problem here is sound matching and fingerprinting are fuzzy and relatively expensive operations.
Ideally I want to them as few times as possible. Ideas?
Sounds like you're not looking to spend a lot of time delving into audio processing/engineering, and hence you want something you can quickly understand and just works. If you're willing to go with something more complex see here for a very good reference.
That being the case, I'd expect simple loudness and zero crossing measures would be sufficient to identify portions of sound. This is great because you can use techniques similar to rsync.
Choose some number of samples as a chunk size and march through your reference audio data at a regular interval. (Let's call it 'chunk size'.) Calculate the zero-crossing measure (you likely want a logarithm (or a fast approximation) of a simple zero-crossing count). Store the chunks in a 2D spatial structure based on time and the zero-crossing measure.
Then march through your actual audio data a much finer step at a time. (Probably doesn't need to be as small as one sample.) Note that you don't have to recompute the measures for the entire chunk size -- just subtract out the zero-crossings no longer in the chunk and add in the new ones that are. (You'll still need to compute the logarithm or approximation thereof.)
Look for the 'next' chunk with a close enough frequency. Note that since what you're looking for is in order from start to finish, there's no reason to look at -all- chunks. In fact, we don't want to since we're far more likely to get false positives.
If the chunk matches well enough, see if it matches all the way out to silence.
The only concerning point is the 2D spatial structure, but honestly this can be made much easier if you're willing to forgive a strict window of approximation. Then you can just have overlapping bins. That way all you need to do is check two bins for all the values after a certain time -- essentially two binary searches through a search structure.
The disadvantage to all of this is it may require some tweaking to get right and isn't a proven method.
If you can reliably distinguish silence from non-silence as you suggest and if the only differences are insertions of silence, then it seems the only non-trivial case is where silence is inserted where there was none before:
ref: --part1part2--
syn: ---part1---part2----
If you can make your chunk size adaptive to the silence, your algorithm should be fine. That is, if your chunk size is equivalent to two characters in the above example, your algorithm would recognize "pa" matches "pa" and "rt" matches "rt" but for the third chunk it must recognize the silence in syn and adapt the chunk size to compare "1" to "1" instead of "1p" to "1-".
For more complicated edits, you might be able to adapt a weighted Shortest Edit Distance algorithm with removing silence have 0 cost.