Do the ski gloves fit?

Problem: I need to invert a spectrogram to listen to the signal. And I want to do it the best way possible.

Solution: It isn’t all that sophisticated. Here’s a limited implementation of Yang’s ICASSP 2008 solution; which is — surprise surprise — quite nearly the same as the ubiquitous heuristic overlap and add istft.

It’s limited in the following ways:

• The least-squares solution is not implemented. This means that LS solution is available only for regularly spaced frequency bands since in that case LS solution reduces to p=2 solution.
• MATLAB fft/ifft routines are used; and they take care of the P matrix (excess zeros to round to powers of 2).
• The matrix algebra is not exploited — this is a quick and dirty test run.

(more…)

Problem I want to generate a sound with frequency glides.

Solution A frequency glide is a sound whose frequency evolves with time. Once the frequency is characterized, i.e. frequency at any point in time is known, a signal can be generated using a co/sine function.
(more…)

Problem I want to compute cross-correlogram. Fast. Can it not be done quick and dirty in the spectral domain?

Solution Cross-correlation is straightforward to compute in spectral domain as ft(sig1)*conj(ft(sig2)). Computing the cross-correlogram requires successive windowed cross-correlation. MATLAB specgram does just that: can we leverage it?
(more…)

Problem I want to compare two surfaces pixel by pixel. Putting them side by side is distracting and exact comparison is difficut (I’m not overlaying a grid on it). Is there some way I can lay one map translucent over the other?

Briefly Turn surface A into RGB image, and darken pixels based on surface B. The figure shows surface A with points lying in [0, 1]. A mask to highlight points lying above 0.5 is overlaid in the lower figure.
(more…)

Problem I’m forced into windows-land, and MATLAB spits out figures all over the place. I do I swat the gnats?

Solution Dock those pests.

[[code]]DQpmdW5jdGlvbiBmaWc9ZmlndXJlX2RvY2tlZCh2YXJhcmdpbikNCmlmKGlzcGMoKSksDQogICAgICAgIGZpZz1maWd1cmUoJ1dpbmRvd1N0eWxlJywnZG9ja2VkJywnUmVuZGVyZXInLCdwYWludGVycycsdmFyYXJnaW57On0pOw0KZWxzZSwNCiAgICAgICAgZmlnPWZpZ3VyZSgnUmVuZGVyZXInLCAncGFpbnRlcnMnLCB2YXJhcmdpbns6fSk7DQplbmQ7DQo=[[/code]]

Problem I write decent prototypes in MATLAB. How do I include it in a report so it doesn’t look too bad?

Solution The latex listings package does a pretty good job. And you want some heavy customization. TODO.
(more…)

Problem I spent my entire childhood over this research and when I finally get a cool graph, matlab throws in some ugly white blotches in it. Who hates me, and why?

Solution This might help: set(gcf, 'Renderer', 'painters'); or figure('Renderer', 'painters') while beginning on a new figure. MATLAB chooses a renderer each time it displays an image (or prints one), balancing memory size with processor load, so a surface plot may be rendered or saved as a bitmap (even in an eps file!). To ensure that MATLAB renders the plot for the best display, set the renderer to ‘painters.’

Problem MATLAB figures saved as eps or jpg need to be scaled to put on page, but text gets skewed on scaling, and text size is *always* messed up.

Solution A function that takes figure placement constraints in terms of fraction of page width/height and computes appropriate dimensions of image and saves it.

If the image covers half the page width and quarter of the page height, we can place such 2 figures horizontally, and 4 such figures vertically. To save file with such dimensions in eps format, we call:

    savefigure(gcf, 'figures/coolgraph.eps', 4, 2, 'epsc2');

(more…)

(more…)