using stride tricks instead of for loops.

This should significantly speed up energy computation, though I havn't
tested.

Changed the function to return an array is length N-windowSize,
instead of zero padding to N. Is this ok?

Added a simple doctest.

pymir/audio/energy.py

@@ -3,17 +3,38 @@
 
 import numpy
 import scipy
+from numpy.lib import stride_tricks
 
 def energy(signal):
+    """
+    This function does something.
+
+    Example:
+    >>> from test import chirp
+    >>> s = chirp()
+    >>> e = energy(s)
+
+    Below is for testing that I don't break things
+    >>> import hashlib
+    >>> hashlib.md5(e).hexdigest()
+    '615c65bb9e9055a75c32ceb5d309e597'
+    >>> hashlib.sha1(e).hexdigest()
+    '531b9ba53419bec6d4f679adc918e7449e17e452'
+    """
     N = len(signal)
 
     windowSize = 256
     window = numpy.hamming(windowSize)
 
-    e = scipy.zeros(N)
-
-    for i in range(0, N - windowSize):
-        e[i] = numpy.sum(window * numpy.power(signal[i:i + windowSize], 2) ) / windowSize
-
-
-    return e
+    n = N - windowSize #number of windowed samples.
+
+    # Create a view of signal who's shape is (n, windowSize). Use stride_tricks such that each stide jumps only one item.
+    s = stride_tricks.as_strided(signal,shape=(n,windowSize), strides=(signal.itemsize,signal.itemsize))
+    e = (window * numpy.power(s, 2)).sum(1) / windowSize
+    return e
+
+def _test():
+    import doctest
+    doctest.testmod(verbose=True)
+if __name__ == '__main__':
+    _test()