python - Find large number of consecutive values fulfilling condition in a numpy array -

i have audio data loaded in numpy array , wish segment data finding silent parts, i.e. parts audio amplitude below threshold on a period in time.

an extremely simple way this:

values = ''.join(("1" if (abs(x) < silence_threshold) else "0" x in samples)) pattern = re.compile('1{%d,}'%int(min_silence))                                                                            match in pattern.finditer(values):    # code goes here 

the code above finds parts there @ least min_silence consecutive elements smaller silence_threshold.

now, obviously, above code horribly inefficient , terrible abuse of regular expressions. there other method more efficient, still results in equally simple , short code?

here's numpy-based solution.

i think (?) should faster other options. it's clear.

however, require twice memory various generator-based solutions. long can hold single temporary copy of data in memory (for diff), , boolean array of same length data (1-bit-per-element), should pretty efficient...

import numpy np  def main():     # generate random data     x = np.cumsum(np.random.random(1000) - 0.5)     condition = np.abs(x) < 1      # print start , stop indicies of each region absolute      # values of x below 1, , min , max of each of these regions     start, stop in contiguous_regions(condition):         segment = x[start:stop]         print start, stop         print segment.min(), segment.max()  def contiguous_regions(condition):     """finds contiguous true regions of boolean array "condition". returns     2d array first column start index of region ,     second column end index."""      # find indicies of changes in "condition"     d = np.diff(condition)     idx, = d.nonzero()       # need start things after change in "condition". therefore,      # we'll shift index 1 right.     idx += 1      if condition[0]:         # if start of condition true prepend 0         idx = np.r_[0, idx]      if condition[-1]:         # if end of condition true, append length of array         idx = np.r_[idx, condition.size] # edit      # reshape result 2 columns     idx.shape = (-1,2)     return idx  main()