In [38]:

    

filename = "test_data/n6mab031109.dat"
filename_h5 = "test_data/n6mab031109.h5"


    

In [34]:

    

import numpy as np
import tables
import os

datatype = np.dtype(np.int16)
nchannels = 32
freq = 20000.
rowsize_total = os.path.getsize(filename) / (datatype.itemsize * nchannels)
duration_total = rowsize_total / float(freq)
duration = 60.


    

In [36]:

    

X = np.fromfile(filename, dtype=dtype, count=channels * int(duration * freq))
X = X.reshape((-1, channels)).T


    

In [48]:

    

f = tables.openFile(filename_h5, 'w')
atom = tables.Atom.from_dtype(X.dtype)
ds = f.createEArray(f.root, 'raw_data', atom, shape=(nchannels, 0), expectedrows=rowsize_total)
ds.append(X)
f.close()


    

In [49]:

    

f = tables.openFile(filename_h5, 'r')
ds = f.root.raw_data


    

In [50]:

    

%timeit -n 1 -r 1 x = ds[::10,:]


    

    




1 loops, best of 1: 380 ms per loop

In [ ]:

    

f.close()