def get_silence(self, p=0.250, v=150, s=0.):
"""
Estimates silences from an audio file.
@deprecated
@param p (float) Minimum silence duration in seconds
@param v (int) Expected minimum volume (rms value)
@param s (float) Shift delta duration in seconds
@return a set of frames corresponding to silences.
"""
self.channel.seek(0)
self.silence = []
# Once silence has been found, continue searching in this interval
nbreadframes = int(self.volstats.get_winlen() * self.channel.get_framerate())
afterloop_frames = int(nbreadframes/2) #int((nbreadframes/2) * self.channel.get_framerate())
initpos = i = self.channel.tell()
# This scans the file in steps of frames whether a section's volume
# is lower than silence_cap, if it is it is written to silence.
while i < self.channel.get_nframes():
curframe = self.channel.get_frames(nbreadframes)
a = AudioFrames( curframe, self.channel.get_sampwidth(), 1)
#volume = audioutils.get_rms(curframe, self.channel.get_sampwidth())
volume = a.rms()
if volume < v:
# Continue searching in smaller steps whether the silence is
# longer than read frames but smaller than read frames * 2.
while volume < v and self.channel.tell() < self.channel.get_nframes():
curframe = self.channel.get_frames(afterloop_frames)
a = AudioFrames( curframe, self.channel.get_sampwidth(), 1)
volume = a.rms()
#volume = audioutils.get_rms(curframe, self.channel.get_sampwidth())
# If the last sequence of silence ends where the new one starts
# it's a continuous range.
if self.silence and self.silence[-1][1] == i:
self.silence[-1][1] = self.channel.tell()
else:
# append if silence is long enough
duree = self.channel.tell() - i
nbmin = int( (p+s) * self.channel.get_framerate())
if duree > nbmin:
# Adjust silence start-pos
__startpos = i + ( s * self.channel.get_framerate() )
# Adjust silence end-pos
__endpos = self.channel.tell() - ( s * self.channel.get_framerate() )
self.silence.append([__startpos, __endpos])
i = self.channel.tell()
# Return the position in the file to where it was when we got it.
self.channel.seek(initpos)
def __init__(self, channel, winlen=0.01):
"""
Constructor.
@param channel (Channel) The channel to work on.
@param winlen (float) Window length to estimate the volume.
"""
BaseVolume.__init__(self)
self.winlen = winlen
# Remember current position
pos = channel.tell()
# Rewind to the beginning
channel.rewind()
# Constants
nbframes = int(winlen * channel.get_framerate())
nbvols = int(channel.get_duration()/winlen) + 1
self.volumes = [0]*nbvols
for i in range(nbvols):
frames = channel.get_frames( nbframes )
a = AudioFrames( frames, channel.get_sampwidth(), 1)
self.volumes[i] = a.rms()
if self.volumes[-1] == 0:
self.volumes.pop()
# Returns to the position where we was before
channel.seek(pos)
self.rms = channel.rms()
def rms(self):
"""
Return the root mean square of the whole file
@return the root mean square of the audio file
"""
pos = self.tell()
self.seek(0)
a = AudioFrames(self.read_frames(self.get_nframes()), self.get_sampwidth(), self.get_nchannels())
self.seek(pos)
return a.rms()
