Welch's method

<h2 id="definition-and-procedure">Definition and procedure</h2>
<p>The Welch method is based on <a href="/facts/Bartlett%2527s_method/SssaAGxj">Bartlett's method</a> and differs in two ways:
</p>
<ol><li>The signal is split up into overlapping segments: the original data segment is split up into L data segments of length M, overlapping by D points.
<ol><li>If D = M / 2, the overlap is said to be 50%</li>
<li>If D = 0, the overlap is said to be 0%. This is the same situation as in the <a href="/facts/Bartlett%2527s_method/SssaAGxj">Bartlett's method</a>.</li></ol></li>
<li>The overlapping segments are then windowed: After the data is split up into overlapping segments, the individual L data segments have a window applied to them (in the time domain).
<ol><li>Most <a href="/facts/Window_function/5Xj7h2ah">window functions</a> afford more influence to the data at the center of the set than to data at the edges, which represents a loss of information.  To mitigate that loss, the individual data sets are commonly overlapped in time (as in the above step).</li>
<li>The windowing of the segments is what makes the Welch method a "modified" <a href="/facts/Periodogram/JMjH2MU0">periodogram</a>.</li></ol></li></ol>
<p>After doing the above, the <a href="/facts/Periodogram/JMjH2MU0">periodogram</a> is calculated by computing the <a href="/facts/Discrete_Fourier_transform/uK9Tjsbl">discrete Fourier transform</a>, and then computing the squared magnitude of the result, yielding power spectrum estimates for each segment. The individual 
spectrum estimates are then averaged, which reduces the variance of the individual power measurements. The end result is an array of power measurements vs. frequency "bin".
</p>
<h2 id="related-approaches">Related approaches</h2>
<p>Other overlapping windowed Fourier transforms include:
</p>
<ul><li><a href="/facts/Modified_discrete_cosine_transform/jCaDMSNY">Modified discrete cosine transform</a></li>
<li><a href="/facts/Short-time_Fourier_transform/XHo1t2m6">Short-time Fourier transform</a></li></ul>
<h2 id="see-also">See also</h2>
<ul><li><a href="/facts/Fast_Fourier_transform/caT7UUCh">Fast Fourier transform</a></li>
<li><a href="/facts/Power_spectrum/q3Rfs7b0">Power spectrum</a></li>
<li><a href="/facts/Spectral_density_estimation/JefWFBsx">Spectral density estimation</a></li></ul>

<ul><li>Welch, P. D. (1967), <a href="https://www.researchgate.net/profile/Fernando_Schlindwein/post/Analyzing_EEGs_with_FFTs_and_PSDs/attachment/59d64f8879197b80779a8a73/AS:498242888896512@1495801769748/download/Welch_The+use+of+fast+Fourier+transform+for+the+estimation+of+power+spectra_A+method+based+on+time+averaging+over+short%2C+modified+periodograms.pdf">"The use of Fast Fourier Transform for the estimation of power spectra: A method based on time averaging over short, modified periodograms"</a> (PDF), <i>IEEE Transactions on Audio and Electroacoustics</i>, AU-15 (2): 70–73, <a href="/facts/Bibcode_(identifier)/9HtdQSGB">Bibcode</a>:<a href="https://ui.adsabs.harvard.edu/abs/1967ITAE...15...70W">1967ITAE...15...70W</a>, <a href="/facts/Doi_(identifier)/muM9Etpq">doi</a>:<a href="https://doi.org/10.1109%2FTAU.1967.1161901">10.1109/TAU.1967.1161901</a></li>
<li>Oppenheim, Alan V.; Schafer, Ronald W. (1975). <i>Digital signal processing</i>. Englewood Cliffs, N.J.: Prentice-Hall. pp. 548–554. <a href="/facts/ISBN_(identifier)/15AdSPa9">ISBN</a> 0-13-214635-5.</li>
<li>Proakis, John G.; Manolakis, Dimitri G. (1996), <a href="https://archive.org/details/digitalsignalpro00proa/page/910"><i>Digital Signal Processing: Principles, Algorithms and Applications</i></a> (3 ed.), Upper Saddle River, NJ: Prentice-Hall, pp. <a href="https://archive.org/details/digitalsignalpro00proa/page/910">910–913</a>, <a href="/facts/ISBN_(identifier)/15AdSPa9">ISBN</a> 9780133942897, sAcfAQAAIAAJ</li></ul>

Welch's method open-in-new

Welch's method