Note: The OP sent me a private message asking for help with Matlab / Octave. Since we like to keep all tutorials available to the public, I said that I would reply in the thread. Here's my reply to him ...
--------------------------------
Hey, this is going to have to be short because we are going out of town visiting relatives for the day, but here's the general idea ...
1. Use your audio editor to produce a very short snippet of the music. To figure out how long it should be, let's assume you want to produce a picture that is 5000 pixels wide (ie, quite high resolution). Let's further assume that you want each sample to have it's own pixel. Therefore, at a sampling rate of, say, 44 KHz, ie, just under 50 kHz, to get 5000 samples, your snippet only needs to be about 0.1 second long.
2. Open Octave / Matlab and use the "audioread()" function to store the samples that are in this short audio file into a Matlab vector, as shown here:
http://www.mathworks.com/help/matlab/ref/audioread.html
3. Next, you need to plot the data. For this, you call Matlab's "plot()" function:
http://www.mathworks.com/help/matlab/ref/plot.html
4. The "plot()" function can be used in many, many different ways, but it always requires two arguments, for your use, the first, X, is simply a vector containing the sample numbers 1, 2, 3 ... etc.. Filling a vector with consecutive integers is one of the first exercises assigned to students learning Matlab. There are many tutorials on this. Google is your friend. ;-)
5. The second argument required by the plot() function is, Y, the vector of sample values. You already have this since you previously called audioread().
6. Before you do anything fancy with the plot() function (ie, like a transparent background), I suggest that you begin with a very simple plot, a black line graph on a plain white background. This is the default for the plot() function. Once you see something, experiment with the zillion options available in Matlab to get the graph to look just like what you had envisioned.
7. Once you see the graph of the part of the audio waveform that you want (ie, looks visually interesting, contains a continuous line of the right thickness, but without markers for each data point, is the right length, etc.), then you can consider exporting the graph as a PNG or a TIFF with transparency using the export_fig() function:
http://www.mathworks.com/matlabcentral/newsreader/view_thread/170398
http://www.mathworks.com/matlabcentral/fileexchange/23629-export-fig
The above procedure may look more complicated and time consuming than it really is, but the quality of the results and the flexibility it affords is stupendous.
Note: You asked about PhotoZoom. With the above procedure, since the output graph is already a PNG that is many thousand pixels wide and already has a transparent background, you don't need to use PhotoZoom, nor do you need to carefully select the background color and delete it.
HTH,
Tom M