Bootcamp Audio Day

import numpy as np

import pylab as plt

from scikits.audiolab import wavread

#################################################

############ EXTRACT AUDIO FROM FILE ############

#################################################

x, fs, enc = wavread("viola.wav")

#################################################

####### CALCULATE RMS OF EACH AUDIO BLOCK #######

#################################################

hop_size = 512                             # set hop size

frame_size = hop_size*2                     # set frame size

frame = np.zeros(frame_size)                # initialise frame with zeros

window = np.hanning(frame_size)             # create window of the same length as the hop size

# create empty numpy array to hold our 

rms = np.array([])

# run through signal frame by frame 

for n in range(0,x.size-hop_size,hop_size):

    # extract a segment of length hop_size

    frame = x[n:n+hop_size]                               

    # calculate RMS

    rms_val = np.sqrt(np.power(frame,2).mean())

    # add amplitude to our numpy array

    rms = np.append(rms,rms_val)

print rms

""" I AM THE OVERLAP VERSION

# run through signal frame by frame 

for n in range(0,x.size-hop_size,hop_size):

    # extract a segment of length hop_size

    buffer = x[n:n+hop_size]                               

    # add new segment to frame, shifting back samples of frame

    frame = np.append(frame[hop_size:frame_size],buffer)  

    # calculate RMS

    rms_val = np.sqrt(np.power(frame,2).mean())

    # add amplitude to our numpy array

    rms = np.append(rms,rms_val)

print rms

"""

plt.plot(rms)

plt.title("RMS")

plt.xlabel("time")

plt.ylabel("value")

plt.show()