def run(self):
global hitFound
#Initialize and set the properties of PCM object
card = 'default'
audioInput = alsaaudio.PCM(alsaaudio.PCM_CAPTURE, alsaaudio.PCM_NORMAL, card)
audioInput.setchannels(2)
audioInput.setrate(44100)
audioInput.setformat(alsaaudio.PCM_FORMAT_S16_LE)
audioInput.setperiodsize(160)
oldL = 1
oldR = 1
try:
#Start an infite loop that gets and analyzes audio data
count = 0
spike = 0
while True:
l, data = audioInput.read()
if l > 0:
spike = 0
lchan = audioop.tomono(data, 2, 1, 0)
rchan = audioop.tomono(data, 2, 0, 1)
lmax = audioop.max(lchan, 2)
rmax = audioop.max(rchan, 2)
#print str(lmax) + " " + str(rmax)
if spike < lmax:
spike = lmax
if spike < rmax:
spike = rmax
#print spike
if spike > 1000:
hitFound = True
time.sleep(.001) #audio refresh rate
except KeyboardInterrupt :
sys.exit() #TODO make it actually exit
def load_wav(fn, stereo=True):
with wave.open(fn, mode='rb') as audio_file:
params = audio_file.getparams()
print(f'''Audio file "{fn}":
Channels : {params.nchannels}
Framerate : {params.framerate}
Sample width : {params.sampwidth} Bytes
Sample num : {params.nframes}
''')
sample_bytes = audio_file.readframes(params.nframes)
ch_left = audioop.tomono(sample_bytes, params.sampwidth, 1, 0)
if stereo:
ch_right = audioop.tomono(sample_bytes, params.sampwidth, 0, 1)
samples = [(l, r) for l, r in zip(list_samples(ch_left, params),
list_samples(ch_right, params))]
else:
samples = list_samples(ch_left, params)
return samples, params
def play(self,fileName):
# Initialise matrix
matrix=[0,0,0,0,0,0,0,0]
# Set up audio
wavfile = wave.open(fileName,'r')
chunk = 1024
output = aa.PCM(aa.PCM_PLAYBACK, aa.PCM_NORMAL)
output.setchannels(1)
output.setrate(22050)
output.setformat(aa.PCM_FORMAT_S16_LE)
output.setperiodsize(chunk)
data = wavfile.readframes(chunk)
try:
while data!='':
output.write(data)
# Split channel data and find maximum volume
channel_l=audioop.tomono(data, 2, 1.0, 0.0)
channel_r=audioop.tomono(data, 2, 0.0, 1.0)
max_vol_factor =5000
max_l = audioop.max(channel_l,2)/max_vol_factor
max_r = audioop.max(channel_r,2)/max_vol_factor
for i in range (1,8):
self.generateMouthSignal((1<<max_r)-1)
data = wavfile.readframes(chunk)
except:
data = None
os.system( '/etc/init.d/alsa-utils restart' )
sleep( .25 )
def listen():
#Initialize and set the properties of PCM object
card = 'default'
audioInput = alsaaudio.PCM(alsaaudio.PCM_CAPTURE, alsaaudio.PCM_NORMAL, card)
audioInput.setchannels(2)
audioInput.setrate(44100)
audioInput.setformat(alsaaudio.PCM_FORMAT_S16_LE)
audioInput.setperiodsize(160)
oldL = 1
oldR = 1
try:
#Start an big loop that gets and analyzes audio data
for i in range(0, 5000):
l, data = audioInput.read()
if l > 0:
spike = 0
lchan = audioop.tomono(data, 2, 1, 0)
rchan = audioop.tomono(data, 2, 0, 1)
lmax = audioop.max(lchan, 2)
rmax = audioop.max(rchan, 2)
if spike < lmax:
spike = lmax
if spike < rmax:
spike = rmax
if spike > 1000:
startStopTime()
time.sleep(.001) #audio refresh rate
except KeyboardInterrupt :
sys.exit() #TODO make it actually exit
def write_sound(self, path, suffix='.wav', devide_stereo=False) :
i = 0
if devide_stereo == True :
if self.__channels == 2 :
frames_r = []
frames_l = []
for frames in self.__sound_frames :
for frame in frames :
frame_l = audioop.tomono(frame, 2, 1, 0)
frames_l.append(frame_l)
frame_r = audioop.tomono(frame, 2, 0, 1)
frames_r.append(frame_r)
filename = path + str(i)
self.__write_soundfile(frames_r, filename + '_r' + suffix, 1)
self.__write_soundfile(frames_l, filename + '_l' + suffix, 1)
i = i + 1
frames_l.clear()
frames_r.clear()
else :
print('Recording Sound is not Stereo Sound.')
print('So, Write only Monaural Sound')
i = 0
for frame in self.__sound_frames :
filename = path + str(i) + suffix
self.__write_soundfile(frame, filename)
i = i + 1
self.__write_log(path)
return
def __db_level(self, rms_mode: bool = False) -> Tuple[float, float]:
"""
Returns the average audio volume level measured in dB (range -60 db to 0 db)
If the sample is stereo, you get back a tuple: (left_level, right_level)
If the sample is mono, you still get a tuple but both values will be the same.
This method is probably only useful if processed on very short sample fragments in sequence,
so the db levels could be used to show a level meter for the duration of the sample.
"""
maxvalue = 2**(8 * self.__samplewidth - 1)
if self.nchannels == 1:
if rms_mode:
peak_left = peak_right = (audioop.rms(
self.__frames, self.__samplewidth) + 1) / maxvalue
else:
peak_left = peak_right = (audioop.max(
self.__frames, self.__samplewidth) + 1) / maxvalue
else:
left_frames = audioop.tomono(self.__frames, self.__samplewidth, 1,
0)
right_frames = audioop.tomono(self.__frames, self.__samplewidth, 0,
1)
if rms_mode:
peak_left = (audioop.rms(left_frames, self.__samplewidth) +
1) / maxvalue
peak_right = (audioop.rms(right_frames, self.__samplewidth) +
1) / maxvalue
else:
peak_left = (audioop.max(left_frames, self.__samplewidth) +
1) / maxvalue
peak_right = (audioop.max(right_frames, self.__samplewidth) +
1) / maxvalue
# cut off at the bottom at -60 instead of all the way down to -infinity
return max(20.0 * math.log(peak_left, 10),
-60.0), max(20.0 * math.log(peak_right, 10), -60.0)
def __init__(self):
'''AudioStream allows to stream audiodata received from a microphone via ROS'''
self._running = False
# global audio recorder (initialized in startStream)
self._recorder = None
# init ROS
rospy.init_node('audiostream', anonymous=True)
# global publisher variables (initialized in startStream)
self._publisherLeft = None
self._publisherRight= None
# global service variables (startStream initializes the other ones)
self._getWidthSrv = None
self._getRateSrv = None
self._startSrv = rospy.Service('nico/audiostream/startStream', nicomsg.srv.StartAudioStream, self._ROSPY_startStream)
self._stopSrv = None
# actual streaming
sampleIndex = 0
while not rospy.is_shutdown():
if self._running and self._recorder.get_number_of_samples() > sampleIndex:
sample = self._recorder.get_chunk(sampleIndex)
sampleIndex += 1
msg = nicomsg.msg.hs()
msg.header.stamp = rospy.Time.now()
if self._channels[0]:
msg.param1 = audioop.tomono(sample, self._recorder.get_sample_width(), 1, 0)
self._publisherLeft.publish(msg)
if self._channels[1]:
msg.param1 = audioop.tomono(sample, self._recorder.get_sample_width(), 0, 1)
self._publisherRight.publish(msg)
if self._running:
self.stopStream()
def get_recognition_result(self):
print("Convert bytes to 1 channel")
audio_bytes_4channels = b''.join(self.audio_fragment_array)
audio_bytes_2channels = audioop.tomono(audio_bytes_4channels, 2, 0.5,
0.5)
audio_bytes_1channel = audioop.tomono(audio_bytes_2channels, 2, 0.5,
0.5)
print("Start speech interaction request to google")
request_json = self.services.google_handle.speech_to_text_api(
base64.b64encode(audio_bytes_1channel), self.sample_frequency)
dict_output = {"success": False, "results": ()}
if request_json:
print(request_json)
dict_output["success"] = True
recognition_result = str(
request_json['results'][0]['alternatives'][0]['transcript'])
confidence_result = float(
request_json['results'][0]['alternatives'][0]['confidence'])
language_result = request_json['results'][0]['languageCode']
print("Request succeed, result: {}, confidence: {}".format(
recognition_result, confidence_result))
dict_output["results"] = (recognition_result, language_result,
confidence_result)
return dict_output
def __db_level(self, rms_mode=False):
"""
Returns the average audio volume level measured in dB (range -60 db to 0 db)
If the sample is stereo, you get back a tuple: (left_level, right_level)
If the sample is mono, you still get a tuple but both values will be the same.
This method is probably only useful if processed on very short sample fragments in sequence,
so the db levels could be used to show a level meter for the duration of the sample.
"""
maxvalue = 2**(8*self.__samplewidth-1)
if self.nchannels == 1:
if rms_mode:
peak_left = peak_right = (audioop.rms(self.__frames, self.__samplewidth)+1)/maxvalue
else:
peak_left = peak_right = (audioop.max(self.__frames, self.__samplewidth)+1)/maxvalue
else:
left_frames = audioop.tomono(self.__frames, self.__samplewidth, 1, 0)
right_frames = audioop.tomono(self.__frames, self.__samplewidth, 0, 1)
if rms_mode:
peak_left = (audioop.rms(left_frames, self.__samplewidth)+1)/maxvalue
peak_right = (audioop.rms(right_frames, self.__samplewidth)+1)/maxvalue
else:
peak_left = (audioop.max(left_frames, self.__samplewidth)+1)/maxvalue
peak_right = (audioop.max(right_frames, self.__samplewidth)+1)/maxvalue
# cut off at the bottom at -60 instead of all the way down to -infinity
return max(20.0*math.log(peak_left, 10), -60.0), max(20.0*math.log(peak_right, 10), -60.0)
def play(self, fileName):
# Initialise matrix
matrix = [0, 0, 0, 0, 0, 0, 0, 0]
# Set up audio
wavfile = wave.open(fileName, 'r')
chunk = 1024
output = aa.PCM(aa.PCM_PLAYBACK, aa.PCM_NORMAL)
output.setchannels(1)
output.setrate(22050)
output.setformat(aa.PCM_FORMAT_S16_LE)
output.setperiodsize(chunk)
data = wavfile.readframes(chunk)
try:
while data != '':
output.write(data)
# Split channel data and find maximum volume
channel_l = audioop.tomono(data, 2, 1.0, 0.0)
channel_r = audioop.tomono(data, 2, 0.0, 1.0)
max_vol_factor = 5000
max_l = audioop.max(channel_l, 2) / max_vol_factor
max_r = audioop.max(channel_r, 2) / max_vol_factor
for i in range(1, 8):
self.generateMouthSignal((1 << max_r) - 1)
data = wavfile.readframes(chunk)
except:
data = None
os.system('/etc/init.d/alsa-utils restart')
sleep(.25)
def recv() :
global client, inp, etc, trig_this_time, trig_last_time, sin
if aout_jack:
# get audio (with 16 bit signed format)
data_l = 32767 * inp[0].get_array()
data_r = 32767 * inp[1].get_array()
else:
# get audio from alsa
l,data = inp.read()
if l > 0:
ldata = audioop.tomono(data, 2, 1, 0)
rdata = audioop.tomono(data, 2, 0, 1)
peak = 0
# try :
for i in range(0,100) :
avg_l = 0
avg_r = 0
for j in range(3):
if aout_jack:
avg_l += data_l[3 * i + j]
avg_r += data_r[3 * i + j]
else:
if l > 0:
# the following 2 lines crashed the python engine -> optimize alsa settings? probably wrong buffer size?
# avg_l +=audioop.getsample(ldata, 2, (i * 3) + j)
# avg_r +=audioop.getsample(rdata, 2, (i * 3) + j)
# this is not a solution, only a fix
avg_l = avg_l
avg_r = avg_r
# except :
# pass
avg_l = avg_l / 3
avg_r = avg_r / 3
avg = (avg_l + avg_r) / 2
# scale it
avg = int(avg * etc.audio_scale)
avg_l = int(avg_l * etc.audio_scale)
avg_r = int(avg_r * etc.audio_scale)
if (avg > 20000) :
trig_this_time = time.time()
if (trig_this_time - trig_last_time) > .05:
if etc.audio_trig_enable: etc.audio_trig = True
trig_last_time = trig_this_time
if avg > peak :
etc.audio_peak = avg
peak = avg
# if the trigger button is held
if (etc.trig_button) :
etc.audio_in[i] = sin[i]
etc.audio_left[i] = sin[i]
etc.audio_right[i] = sin[i]
else :
etc.audio_in[i] = avg
etc.audio_left[i] = avg_l
etc.audio_right[i] = avg_r
def mul_stereo(fileName, width, lfactor, rfactor):
lsample = audioop.tomono(fileName, width, 1, 0)
rsample = audioop.tomono(fileName, width, 0, 1)
lsample = audioop.mul(lsample, width, lfactor)
rsample = audioop.mul(rsample, width, rfactor)
lsample = audioop.tostereo(lsample, width, 1, 0)
rsample = audioop.tostereo(rsample, width, 0, 1)
return audioop.add(lsample, rsample, width)
def mul_stereo(fileName,width,lfactor,rfactor):
lsample = audioop.tomono(fileName, width, 1, 0)
rsample = audioop.tomono(fileName,width, 0, 1)
lsample = audioop.mul(lsample,width,lfactor)
rsample = audioop.mul(rsample, width,rfactor)
lsample = audioop.tostereo(lsample, width, 1, 0)
rsample = audioop.tostereo(rsample, width, 0, 1)
return audioop.add(lsample,rsample,width)
def get_dr(filename, floats=False):
with wave.open(filename, "rb") as f:
channels = f.getnchannels()
if channels not in (1, 2):
# TODO unpack n channels
raise NotImplementedError(
"We only handle mono or stereo at the moment")
framesize = f.getsampwidth()
if framesize != 2:
# TODO map framesize to struct module constants
raise NotImplementedError(
"We only handle 16 bit formats at the moment")
framerate = f.getframerate()
total = f.getnframes()
read = 0
peaks = [[] for i in range(channels)]
rmss = [[] for i in range(channels)]
while True:
# read three seconds of data
block = f.readframes(framerate * 3)
expected = framerate * 3 * channels * framesize
if len(block) < expected:
# EOF
break
read += 3 * framerate
# unpack
if channels == 2:
chansamples = [
audioop.tomono(block, framesize, 1, 0),
audioop.tomono(block, framesize, 0, 1)
]
else:
chansamples = [block]
for i, chan in enumerate(chansamples):
peak = audioop.max(chan, framesize) / NORM
rms = math.sqrt(2) * audioop.rms(chan, framesize) / NORM
peaks[i].append(peak)
rmss[i].append(rms)
drs = []
for c in range(channels):
peaks[c].sort()
rmss[c].sort()
p2 = peaks[c][-2]
if p2 == 0:
raise SilentTrackError
N = int(0.2 * len(peaks[c]))
if N == 0:
raise TooShortError
r = math.sqrt(sum(i**2 for i in rmss[c][-N:]) / N)
dr = -to_db(r / p2)
drs.append(dr)
if not floats:
fdr = round(sum(drs) / len(drs))
else:
fdr = sum(drs) / len(drs)
return fdr
def callback(in_data, frame_count, time_info, status):
if self.state == self.RECORDING:
self._wavefile.writeframes(in_data)
self._time_info = time_info
lsample = audioop.tomono(in_data, self._sample_width, 1, 0)
rsample = audioop.tomono(in_data, self._sample_width, 0, 1)
l_max = audioop.max(lsample, self._sample_width)/self._int_max
r_max = audioop.max(rsample, self._sample_width)/self._int_max
self.levelUpdated.emit([l_max, r_max])
return (in_data, pyaudio.paContinue)
def recv():
global client, inp, etc, trig_this_time, trig_last_time, sin
if aout_norns:
# get audio (with 16 bit signed format)
data_l = 32767 * inp[0].get_array()
data_r = 32767 * inp[1].get_array()
else:
# get audio from alsa
l, data = inp.read()
if l > 0:
ldata = audioop.tomono(data, 2, 1, 0)
rdata = audioop.tomono(data, 2, 0, 1)
peak = 0
# try :
for i in range(0, 100):
avg_l = 0
avg_r = 0
for j in range(3):
if aout_norns:
avg_l += data_l[3 * i + j]
avg_r += data_r[3 * i + j]
else:
if l > 0:
avg_l += audioop.getsample(ldata, 2, (i * 3) + j)
avg_r += audioop.getsample(rdata, 2, (i * 3) + j)
# except :
# pass
avg_l = avg_l / 3
avg_r = avg_r / 3
avg = (avg_l + avg_r) / 2
# scale it
avg = int(avg * etc.audio_scale)
avg_l = int(avg_l * etc.audio_scale)
avg_r = int(avg_r * etc.audio_scale)
if (avg > 20000):
trig_this_time = time.time()
if (trig_this_time - trig_last_time) > .05:
if etc.audio_trig_enable: etc.audio_trig = True
trig_last_time = trig_this_time
if avg > peak:
etc.audio_peak = avg
peak = avg
# if the trigger button is held
if (etc.trig_button):
etc.audio_in[i] = sin[i]
etc.audio_left[i] = sin[i]
etc.audio_right[i] = sin[i]
else:
etc.audio_in[i] = avg
etc.audio_left[i] = avg_l
etc.audio_right[i] = avg_r
def pan(slice, pan_pos=0.5, amp=1.0):
amps = pantamp(pan_pos)
lslice = audioop.tomono(slice, audio_params[1], 1, 0)
lslice = audioop.tostereo(lslice, audio_params[1], amps[0], 0)
rslice = audioop.tomono(slice, audio_params[1], 0, 1)
rslice = audioop.tostereo(rslice, audio_params[1], 0, amps[1])
slice = audioop.add(lslice, rslice, audio_params[1])
return audioop.mul(slice, audio_params[1], amp)
def test_tomono(self):
for w in 1, 2, 4:
data1 = datas[w]
data2 = bytearray(2 * len(data1))
for k in range(w):
data2[k :: 2 * w] = data1[k::w]
self.assertEqual(audioop.tomono(data2, w, 1, 0), data1)
self.assertEqual(audioop.tomono(data2, w, 0, 1), b"\0" * len(data1))
for k in range(w):
data2[k + w :: 2 * w] = data1[k::w]
self.assertEqual(audioop.tomono(data2, w, 0.5, 0.5), data1)
def test_tomono(self):
for w in 1, 2, 4:
data1 = datas[w]
data2 = bytearray(2 * len(data1))
for k in range(w):
data2[k::2*w] = data1[k::w]
self.assertEqual(audioop.tomono(str(data2), w, 1, 0), data1)
self.assertEqual(audioop.tomono(str(data2), w, 0, 1), b'\0' * len(data1))
for k in range(w):
data2[k+w::2*w] = data1[k::w]
self.assertEqual(audioop.tomono(str(data2), w, 0.5, 0.5), data1)
def db_level(data, samplewidth=2, rms_mode=False):
maxvalue = 2**(8 * samplewidth - 1)
left_frames = audioop.tomono(data, samplewidth, 1, 0)
right_frames = audioop.tomono(data, samplewidth, 0, 1)
if rms_mode:
peak_left = (audioop.rms(left_frames, samplewidth) + 1) / maxvalue
peak_right = (audioop.rms(right_frames, samplewidth) + 1) / maxvalue
else:
peak_left = (audioop.max(left_frames, samplewidth) + 1) / maxvalue
peak_right = (audioop.max(right_frames, samplewidth) + 1) / maxvalue
return peak_left * 1000, peak_right * 1000
def _convert_data(self, data: bytes, to_depth: int, to_channels: int,
to_rate: int, to_unsigned: bool = False) -> bytes:
"""Convert audio data."""
out_width = to_depth // 8
if self._from_float:
ldata = audioop.tomono(data, self._width, 1, 0)
rdata = audioop.tomono(data, self._width, 0, 1)
for mono_data in [ldata, rdata]:
float_array = array('f', mono_data)
out_array = array('i' if self._out_depth > 16 else 'h')
for i in float_array:
if i > 1.0:
i = 1.0
elif i < -1.0:
i = -1.0
out_array.append(round(i * 32767.0))
mono_data = out_array.tobytes()
ldata = audioop.tostereo(ldata, self._width, 1, 0)
rdata = audioop.tostereo(rdata, self._width, 0, 1)
data = audioop.add(ldata, rdata, self._width)
if self._to_alaw:
data = audioop.lin2alaw(data, self._width)
if self._depth != to_depth:
data = audioop.lin2lin(
data,
self._width,
out_width
)
if self._unsigned != to_unsigned:
data = audioop.bias(data, out_width, 128)
# Make it stereo
if self._channels < to_channels:
data = audioop.tostereo(data, out_width, 1, 1)
# Make it mono
elif self._channels > to_channels:
data = audioop.tomono(data, out_width, 1, 1)
# Convert the sample rate of the data to the requested rate.
if self._rate != to_rate and data:
data, self._state = audioop.ratecv(
data,
out_width,
to_channels,
self._rate,
to_rate,
self._state,
)
return data
def pan(slice, pan_pos=0.5, amp=1.0):
amps = pantamp(pan_pos)
lslice = audioop.tomono(slice, audio_params[1], 1, 0)
lslice = audioop.tostereo(lslice, audio_params[1], amps[0], 0)
rslice = audioop.tomono(slice, audio_params[1], 0, 1)
rslice = audioop.tostereo(rslice, audio_params[1], 0, amps[1])
slice = audioop.add(lslice, rslice, audio_params[1])
return audioop.mul(slice, audio_params[1], amp)
def get_sound():
data = stream.read(frames)
lsampe = audioop.tomono(data, 2, 1, 0)
rsample = audioop.tomono(data, 2, 0, 1)
temp_l = np.frombuffer(lsampe, np.int16).astype(np.float)
temp_r = np.frombuffer(rsample, np.int16).astype(np.float)
lrms = np.sqrt((temp_l * temp_l).sum() / len(temp_l))
rrms = np.sqrt((temp_r * temp_r).sum() / len(temp_r))
return (lrms, rrms)
def recv() :
global inp, etc, trig_this_time, trig_last_time, sin
# get audio
l,data = inp.read()
peak = 0
while l:
try :
ldata = audioop.tomono(data, 2, 1, 0)
rdata = audioop.tomono(data, 2, 0, 1)
for i in range(0,100) :
avg_l = 0
avg_r = 0
idx = i * 3
avg_l +=audioop.getsample(ldata,2,idx)
avg_r +=audioop.getsample(rdata,2,idx)
idx += 1
avg_l +=audioop.getsample(ldata,2,idx)
avg_r +=audioop.getsample(rdata,2,idx)
idx += 1
avg_l +=audioop.getsample(ldata,2,idx)
avg_r +=audioop.getsample(rdata,2,idx)
avg_l = avg_l / 3
avg_r = avg_r / 3
avg = (avg_l + avg_r) / 2
if (avg > 20000) :
trig_this_time = time.time()
if (trig_this_time - trig_last_time) > .05:
etc.audio_trig = True
trig_last_time = trig_this_time
if avg > peak :
etc.audio_peak = avg
peak = avg
# if the trigger button is held
if (etc.trig_button) :
etc.audio_in[i] = sin[i]
etc.audio_left[i] = sin[i]
etc.audio_right[i] = sin[i]
else :
etc.audio_in[i] = avg
etc.audio_left[i] = avg_l
etc.audio_right[i] = avg_r
except :
pass
l,data = inp.read()
def convert_to(self,
data: bytes,
to_depth: int,
to_channels: int,
to_rate: int,
to_unsigned: bool = False) -> bytes:
"""Convert audio data."""
dest_width = to_depth // 8
print(to_depth, self._depth)
if self._depth != to_depth:
if self._depth == 8:
data = audioop.bias(data, 1, 128)
data = audioop.lin2lin(data, self._width, dest_width)
if to_depth == 8:
data = audioop(data, 1, 128)
if self._unsigned != to_unsigned:
data = audioop.bias(data, dest_width, 128)
# Make it stereo
if self._channels < to_channels:
data = audioop.tostereo(data, dest_width, 1, 1)
# Make it mono
elif self._channels > to_channels:
data = audioop.tomono(data, dest_width, 1, 1)
# print(dest_width)
# Convert the sample rate of the data to the requested rate.
if self._rate != to_rate and data:
data, self._state = audioop.ratecv(data, dest_width, to_channels,
self._rate, to_rate,
self._state, 2, 1)
return data
def read(self, size=-1):
buffer = self.wav_reader.readframes(self.wav_reader.getnframes() if size == -1 else size)
if self.wav_reader.getnchannels() != 1: # stereo audio
buffer = audioop.tomono(
buffer, self.wav_reader.getsampwidth(), 1, 1
) # convert stereo audio data to mono
return buffer
def raw_read(self):
"""Return some amount of data as a raw audio string"""
buf = self.source.raw_read()
if buf is None:
self.eof = True
return None
# Convert channels as needed
if self.set_channels and self.source.channels() != self.set_channels:
if self.set_channels == 1:
buf = audioop.tomono(buf, self.source.raw_width(), .5, .5)
else:
buf = audioop.tostereo(buf, self.source.raw_width(), 1, 1)
# Convert sampling rate as needed
if self.set_sampling_rate and self.source.sampling_rate() != self.set_sampling_rate:
(buf, self.ratecv_state) = audioop.ratecv(buf, self.source.raw_width(), self.channels(), self.source.sampling_rate(), self.set_sampling_rate, self.ratecv_state)
if self.set_raw_width and self.source.raw_width() != self.set_raw_width:
if self.source.raw_width() == 1 and self.source.has_unsigned_singles():
buf = audioop.bias(buf, 1, -128)
buf = audioop.lin2lin(buf, self.source.raw_width(), self.set_raw_width)
if self.set_raw_width == 1 and self.source.has_unsigned_singles():
buf = audioop.bias(buf, 1, 128)
return buf
def openwavfile(self, filename): # open wave file try: infile = wave.open(filename, "r") except wave.Error as e: dialog = gtk.MessageDialog(self.window, gtk.DIALOG_DESTROY_WITH_PARENT, gtk.MESSAGE_ERROR, gtk.BUTTONS_CLOSE, "Error: %s" % e) dialog.run() dialog.destroy() return # get raw frame data (binary string of integers) frames = infile.readframes(infile.getnframes()) # mix down to mono if it's stereo if infile.getnchannels() > 1: frames = audioop.tomono(frames, infile.getsampwidth(), 0.5, 0.5) # get list of integer samples samples = binary_unsigned_integers_to_list(frames, infile.getsampwidth()) # trigger the waveform setup etc self.newaudio(Audio(samples, infile.getframerate(), infile.getsampwidth(), filename))
def to_mono(self):
'Convert stereo audio to mono'
if self.params.nchannels == 1: return self
return self.__class__(self.params,
audioop.tomono(self.frames,
self.params.sampwidth, 0.5, 0.5),
nchannels=1)
def coerce_lin(source_aiff, template_obj):
'''Read data from source, and convert it to match template's params.'''
import audioop
frag = source_aiff.read_lin()
Ss = source_aiff.stream
St = template_obj.stream
# Sample width
if Ss.getsampwidth() != St.getsampwidth():
print 'coerce sampwidth %i -> %i' %(Ss.getsampwidth(), St.getsampwidth())
frag = audioop.lin2lin(frag, Ss.getsampwidth(), St.getsampwidth())
width = St.getsampwidth()
# Channels
if Ss.getnchannels() != St.getnchannels():
print 'coerce nchannels %i -> %i' %(Ss.getnchannels(), St.getnchannels())
if Ss.getnchannels()==2 and St.getnchannels()==1:
frag = audioop.tomono(frag, width, 0.5, 0.5)
elif Ss.getnchannels()==1 and St.getnchannels()==2:
frag = audioop.tostereo(frag, width, 1.0, 1.0)
else:
print "Err: can't match channels"
# Frame rate
if Ss.getframerate() != St.getframerate():
print 'coerce framerate %i -> %i' %(Ss.getframerate(), St.getframerate())
frag,state = audioop.ratecv(
frag, width,
St.getnchannels(),
Ss.getframerate(), # in rate
St.getframerate(), # out rate
None, 2,1
)
return frag
def __init__(self, fname, newrate=0):
"""Initializes an audio file from an uncompressed wavefile on disk.
The file is converted to mono, and if newrate is positive, then the rate is converted"""
import wave, audioop
try: # see if we have numpy
from numpy import array
self.numpy = 1
except ImportError:
self.numpy = 0
# read data
f = wave.open(fname, "rb")
nchans, w, fps, nframes, comptype, compname = f.getparams()
print "Read audio file %s with %d chans, %d width, %d fps and %d frames" % (fname, nchans, w, fps, nframes)
self.width, self.fps = w, fps
self.dat = f.readframes(nframes)
print " Original data length was %d" % (len(self.dat))
# convert to mono and (optionally) convert the rate
self.dat = audioop.tomono(self.dat, w, 0.5, 0.5)
print " After mono, data length is %d" % (len(self.dat))
if newrate > 0:
self.dat, junk = audioop.ratecv(self.dat, w, 1, fps, newrate, None)
self.fps = newrate
print " Converted to new rate %s, and data length is now %d" % (self.fps, len(self.dat))
# now extract the data into a simple array
from audioop import getsample
self.dat = [abs(getsample(self.dat, w, i)) for i in range(len(self.dat) // w)]
print " Final data length is now of length %s" % (len(self.dat),)
if self.numpy:
self.dat = array(self.dat)
def downsampleWav(data, dst, inrate=sampleRate, outrate=16000, inchannels=1, outchannels=1):
try:
s_write = wave.open(dst, 'wb')
except:
print('Failed to open files!')
return False
try:
converted = audioop.ratecv(data, 2, inchannels, inrate, outrate, None)
if outchannels == 1 and inchannels != 1:
converted = audioop.tomono(converted[0], 2, 1, 0)
except:
print('Failed to downsample wav')
return False
try:
s_write.setparams((outchannels, 2, outrate, 0, 'NONE', 'Uncompressed'))
s_write.writeframes(converted[0])
except Exception as e:
print(e)
print('Failed to write wav')
return False
try:
s_write.close()
except:
print('Failed to close wav files')
return False
return True
def calc_volume(wav, vol_rate=DEFAULT_VOL_RATE):
'''
ボリュームデータを計算する。
二乗平均平方根（Root Mean Square）を計算。
'''
w = wav.getsampwidth()
rate = wav.getframerate()
buf = wav.readframes(wav.getnframes())
if wav.getnchannels() == 2:
buf = audioop.tomono(buf, w, 0.5, 0.5)
vol_nframes = wav.getnframes() * vol_rate / rate
step = len(buf) / vol_nframes
step = step + (step % w)
vol = []
for i in range(vol_nframes):
sec = float(i) / vol_rate
wav_f = int(sec * rate)
st = wav_f * w
ed = st + step
rms = audioop.rms(buf[st: ed], w)
vol.append(rms)
return vol
def normalize(self) -> 'Sample':
"""
Normalize the sample, meaning: convert it to the default samplerate, sample width and number of channels.
When mixing samples, they should all have the same properties, and this method is ideal to make sure of that.
"""
if self.__locked:
raise RuntimeError("cannot modify a locked sample")
self.resample(params.norm_samplerate)
if self.samplewidth != params.norm_samplewidth:
# Convert to desired sample size.
self.__frames = audioop.lin2lin(self.__frames, self.samplewidth,
params.norm_samplewidth)
self.__samplewidth = params.norm_samplewidth
if params.norm_nchannels not in (1, 2):
raise ValueError(
"norm_nchannels has invalid value, can only be 1 or 2")
if self.nchannels == 1 and params.norm_nchannels == 2:
# convert to stereo
self.__frames = audioop.tostereo(self.__frames, self.samplewidth,
1, 1)
self.__nchannels = 2
elif self.nchannels == 2 and params.norm_nchannels == 1:
# convert to mono
self.__frames = audioop.tomono(self.__frames, self.__samplewidth,
1, 1)
self.__nchannels = 1
return self
def encode(self,
frame: Frame,
force_keyframe: bool = False) -> Tuple[List[bytes], int]:
assert isinstance(frame, AudioFrame)
assert frame.format.name == "s16"
assert frame.layout.name in ["mono", "stereo"]
channels = len(frame.layout.channels)
data = bytes(frame.planes[0])
timestamp = frame.pts
# resample at 8 kHz
if frame.sample_rate != SAMPLE_RATE:
data, self.rate_state = audioop.ratecv(
data,
SAMPLE_WIDTH,
channels,
frame.sample_rate,
SAMPLE_RATE,
self.rate_state,
)
timestamp = (timestamp * SAMPLE_RATE) // frame.sample_rate
# convert to mono
if channels == 2:
data = audioop.tomono(data, SAMPLE_WIDTH, 1, 1)
data = self._convert(data, SAMPLE_WIDTH)
return [data], timestamp
def splitStereoAudio(path, fn, outputPath=None):
if outputPath is None:
outputPath = join(path, "split_audio")
if not os.path.exists(outputPath):
os.mkdir(outputPath)
name = os.path.splitext(fn)[0]
fnFullPath = join(path, fn)
leftOutputFN = join(outputPath, "%s_L.wav" % name)
rightOutputFN = join(outputPath, "%s_R.wav" % name)
audiofile = wave.open(fnFullPath, "r")
params = audiofile.getparams()
sampwidth = params[1]
nframes = params[3]
audioFrames = audiofile.readframes(nframes)
for leftFactor, rightFactor, outputFN in ((1, 0, leftOutputFN),
(0, 1, rightOutputFN)):
monoAudioFrames = audioop.tomono(audioFrames, sampwidth,
leftFactor, rightFactor)
params = tuple([1, ] + list(params[1:]))
outputAudiofile = wave.open(outputFN, "w")
outputAudiofile.setparams(params)
outputAudiofile.writeframes(monoAudioFrames)
def read(self, size=-1):
#@noter：崔冰
#@description: 将待处理的音频文件的字节序和位数进行相应处理
buffer = self.audio_reader.readframes(
self.audio_reader.getnframes() if size == -1 else size)
if not isinstance(buffer, bytes):
buffer = b"" # 解决方法 https://bugs.python.org/issue24608
sample_width = self.audio_reader.getsampwidth()
if not self.little_endian: # 大端字节序格式转换成小端字节序
if hasattr(
audioop, "byteswap"
): # ``audioop.byteswap``只在Python 3.4 中有(python3.4+ 支持24位)
buffer = audioop.byteswap(buffer, sample_width)
else: # 手动反转每个样本的字节数，但速度较慢
buffer = buffer[sample_width - 1::-1] + b"".join(
buffer[i + sample_width:i:-1]
for i in range(sample_width -
1, len(buffer), sample_width))
# 解决方法 https://bugs.python.org/issue12866
if self.samples_24_bit_pretending_to_be_32_bit: # 我们需要将样本从24位转换为32位，然后才能使用“audioop”功能处理
buffer = b"".join(b"\x00" + buffer[i:i + sample_width]
for i in range(0, len(buffer), sample_width)
) # 由于我们处于小端字节序，因此我们在每个24位采样前添加一个零字节以获取32位采样
sample_width = 4 # 确保我们将缓冲区转换为32位音频，然后将其转换为24位音频
if self.audio_reader.getnchannels() != 1: # 立体声音频
buffer = audioop.tomono(buffer, sample_width, 1,
1) # 将立体声音频转换成单声道
return buffer
def to_mono(self):
if self.channels == 1:
return self
elif self.channels == 2:
return Audio(channels=1, width=self.width, rate=self.rate, data=audioop.tomono(self.data, self.width, 0.5, 0.5))
else:
raise ValueError(f"Can't convert audio with channels={self.channels}")
def read(self, size=-1):
buffer = self.audio_reader.readframes(
self.audio_reader.getnframes() if size == -1 else size)
if not isinstance(buffer, bytes):
buffer = b"" # workaround for https://bugs.python.org/issue24608
sample_width = self.audio_reader.getsampwidth()
if not self.little_endian: # big endian format, convert to little endian on the fly
if hasattr(
audioop, "byteswap"
): # ``audioop.byteswap`` was only added in Python 3.4 (incidentally, that also means that we don't need to worry about 24-bit audio being unsupported, since Python 3.4+ always has that functionality)
buffer = audioop.byteswap(buffer, sample_width)
else: # manually reverse the bytes of each sample, which is slower but works well enough as a fallback
buffer = buffer[sample_width - 1::-1] + b"".join(
buffer[i + sample_width:i:-1]
for i in range(sample_width -
1, len(buffer), sample_width))
# workaround for https://bugs.python.org/issue12866
if self.samples_24_bit_pretending_to_be_32_bit: # we need to convert samples from 24-bit to 32-bit before we can process them with ``audioop`` functions
buffer = b"".join(
"\x00" + buffer[i:i + sample_width]
for i in range(0, len(buffer), sample_width)
) # since we're in little endian, we prepend a zero byte to each 24-bit sample to get a 32-bit sample
if self.audio_reader.getnchannels() != 1: # stereo audio
buffer = audioop.tomono(buffer, sample_width, 1,
1) # convert stereo audio data to mono
return buffer
def downsampleWav(src, dst, inrate=44100, outrate=22050, inchannels=1, outchannels=1):
if not os.path.exists(src):
print 'Source not found!'
return False
s_read = wave.open(src, 'r')
s_write = wave.open(dst, 'w')
n_frames = s_read.getnframes()
data = s_read.readframes(n_frames)
try:
converted = audioop.ratecv(data, 2, inchannels, inrate, outrate, None)
if outchannels == 1:
converted = audioop.tomono(converted[0], 2, 1, 0)
except:
print 'Failed to downsample wav'
return False
try:
s_write.setparams((outchannels, 2, outrate, 0, 'NONE', 'Uncompressed'))
s_write.writeframes(converted)
except:
print 'Failed to write wav'
return False
try:
s_read.close()
s_write.close()
except:
print 'Failed to close wav files'
return False
return True
def testtomono(data): data2 = '' for d in data[0]: data2 = data2 + d + d if audioop.tomono(data2, 1, 0.5, 0.5) <> data[0]: return 0 return 1
def calculate_volume(sphfile,leftch,rightch,SIR):
# compute volume level from sph header.
# Useful to compute the signal-to-interference
# level of stereo sph files.
with open(sphfile) as s:
bytes = s.read()
s1_bytes1 = audioop.tomono(bytes,2,leftch,rightch)
s2_bytes1 = audioop.tomono(bytes,2,rightch,leftch)
s1_bytes = s1_bytes1[1024:]
s2_bytes = s2_bytes1[1024:]
e1 = audioop.rms(s1_bytes,2)*1.0 # make float by multiplying by 1.0
e2 = audioop.rms(s2_bytes,2)*1.0
print e1,e2
vol = math.exp(-1.0*float(SIR)/10)*e1/e2
return vol
def write(self, audio):
if self._user_resample:
# The user can also specify to have ULAW encoded source audio
# converted to linear encoding upon being written.
if self._user_resample.ulaw2lin:
# Possibly skip downsampling if this was triggered, as
# while ULAW encoded audio can be sampled at rates other
# than 8KHz, since this is telephony related, it's unlikely.
audio = audioop.ulaw2lin(audio, 2)
# If the audio isn't already sampled at 8KHz,
# then it needs to be downsampled first
if self._user_resample.rate != 8000:
audio, self._user_resample.ratecv_state = audioop.ratecv(
audio,
2,
self._user_resample.channels,
self._user_resample.rate,
8000,
self._user_resample.ratecv_state,
)
# If the audio isn't already in mono, then
# it needs to be downmixed as well
if self._user_resample.channels == 2:
audio = audioop.tomono(audio, 2, 1, 1)
self._tx_q.put(audio)
def splitStereoAudio(path, fn, outputPath=None):
if outputPath is None:
outputPath = join(path, "split_audio")
if not os.path.exists(outputPath):
os.mkdir(outputPath)
name = os.path.splitext(fn)[0]
fnFullPath = join(path, fn)
leftOutputFN = join(outputPath, "%s_L.wav" % name)
rightOutputFN = join(outputPath, "%s_R.wav" % name)
audiofile = wave.open(fnFullPath, "r")
params = audiofile.getparams()
sampwidth = params[1]
nframes = params[3]
audioFrames = audiofile.readframes(nframes)
for leftFactor, rightFactor, outputFN in ((1, 0, leftOutputFN),
(0, 1, rightOutputFN)):
monoAudioFrames = audioop.tomono(audioFrames, sampwidth, leftFactor,
rightFactor)
params = tuple([
1,
] + list(params[1:]))
outputAudiofile = wave.open(outputFN, "w")
outputAudiofile.setparams(params)
outputAudiofile.writeframes(monoAudioFrames)
def _play(self, data, rate=16000, channels=1, width=2, spectrum=True):
stream = self.pyaudio_instance.open(
format=self.pyaudio_instance.get_format_from_width(width),
channels=channels,
rate=rate,
output=True,
# output_device_index=1,
frames_per_buffer=CHUNK_SIZE,
)
if isinstance(data, types.GeneratorType):
for d in data:
if self.stop_event.is_set():
break
stream.write(d)
if spectrum:
if channels == 2:
d = audioop.tomono(d, 2, 0.5, 0.5)
self.queue.put(d)
else:
stream.write(data)
stream.close()
def getLevel(in_data, leftChannel=True):
normalize = 32767.0
leftFactor = 1 if leftChannel else 0
rightFactor = 1 if not leftChannel else 0
data = audioop.tomono(in_data, 2, leftFactor, rightFactor)
amplitudel = ((audioop.max(data, 2))/normalize)
level = (int(41+(20*(math.log10(amplitudel+(1e-40))))))
return level
def _push_up_some_data(self):
if self.readdev is None:
return
(l, data,) = self.readdev.read()
if self.readchannels == 2:
data = audioop.tomono(data, 2, 1, 1)
if self.encoder and data:
self.encoder.handle_audio(data)
def findfit(scratch_frag, final_frag, sound_file):
'''Calculates the offset (in seconds) between scratch_frag & final_frag.
Both fragments are assumed to contain the same, loud "clapper" event.
The SoundFile object is used for common stream parameters.'''
import audioop
nchannels = sound_file.stream.getnchannels()
framerate = sound_file.stream.getframerate()
width = sound_file.stream.getsampwidth()
assert(width==2)
# Simplify the sound streams to make it quicker to find a match.
# Left channel only.
if nchannels > 1:
scratch_frag_ = audioop.tomono(scratch_frag, width, 1, 0)
final_frag_ = audioop.tomono(final_frag, width, 1, 0)
else:
scratch_frag_ = scratch_frag
final_frag_ = final_frag
nchannels_ = 1
# Downsample to 8000/sec
framerate_ = 8000
scratch_frag_,state =\
audioop.ratecv(scratch_frag_, width, nchannels_, framerate, framerate_, None)
final_frag_,state =\
audioop.ratecv(final_frag_, width, nchannels_, framerate, framerate_, None)
bytes_per_second_ = nchannels_ * framerate_ * width
# Find the clapper in final
length_samples = int(0.001 * framerate * nchannels_) # 0.1 sec
final_off_samples = audioop.findmax(final_frag_, length_samples)
# Search for a 2 second 'needle' centred on where we found the 'clapper'
needle_bytes = 2 * bytes_per_second_
b0 = max(0, final_off_samples * width - int(needle_bytes/2))
print '"clapper" at final:', 1.0*b0/bytes_per_second_, 'sec'
b1 = b0 + needle_bytes
final_clapper_frag = final_frag_[b0:b1]
scratch_off_samples,factor = audioop.findfit(scratch_frag_, final_clapper_frag)
scratch_off_bytes = scratch_off_samples * width
print 'match at scratch:', 1.0*scratch_off_bytes/bytes_per_second_, 'sec', " factor =",factor
# Calculate the offset (shift) between the two fragments.
shift_sec = (scratch_off_bytes - b0) * 1.0 / bytes_per_second_
print 'shift =', shift_sec, 'seconds'
return shift_sec
def testtomono(data):
if verbose:
print 'tomono'
data2 = ''
for d in data[0]:
data2 = data2 + d + d
if audioop.tomono(data2, 1, 0.5, 0.5) <> data[0]:
return 0
return 1
def read_audio( filename, stereo=True ):
"""reads a stereo audio file but returns a mono buffer"""
wfile = wave.open( filename, 'r' )
buf = wfile.readframes( wfile.getnframes() )
# below, parameters are ( buffer, width, lfactor, rfactor)
if stereo:
return audioop.tomono( buf, 2, 1, 0 )
else:
return buf
def _push_up_some_data(self):
from audioop import tomono
try:
data = self.dev.read(320*self._channels)
except IOError:
return None
if self._channels == 2:
data = tomono(data, 2, 1, 1)
if self.encoder and data:
self.encoder.handle_audio(data)
def mono(self, left_factor=1.0, right_factor=1.0):
"""Make the sample mono (1-channel) applying the given left/right channel factors when downmixing"""
assert not self.__locked
if self.__nchannels == 1:
return self
if self.__nchannels == 2:
self.__frames = audioop.tomono(self.__frames, self.__samplewidth, left_factor, right_factor)
self.__nchannels = 1
return self
raise ValueError("sample must be stereo or mono already")
def split_channels(adc_data):
"""
Args:
adc_data: Bunch with the following fields:
- data (binary string or list of binary strings): raw ADC data
- sample_width (int): number of bytes per sample
Returns:
A Bunch with fields:
- voltage (binary string): raw ADC data
- current (binary string): raw ADC data
- sample_width (int): number of bytes per sample
"""
data = Bunch()
data.voltage = audioop.tomono(adc_data.data, adc_data.sample_width, 1, 0)
data.current = audioop.tomono(adc_data.data, adc_data.sample_width, 0, 1)
data.sample_width = adc_data.sample_width
return data
def extractSubwav(fn, outputFN, startT, endT, singleChannelFlag, soxPath=None):
if soxPath is None:
soxPath = "sox" # Assumes it is in the user's path
path, fnNoPath = os.path.split(fn)
resampledAudioPath = join(path, "resampledAudio")
resampledFN = join(resampledAudioPath, fnNoPath)
audiofile = wave.open(fn, "r")
params = audiofile.getparams()
nchannels = params[0]
sampwidth = params[1]
framerate = params[2]
comptype = params[4]
compname = params[5]
# If you are not using the default Julius training model, you might
# need to change the value here. Results will fail if the sampling
# frequency is different.
if framerate != DEFAULT_FRAMERATE:
if not os.path.exists(resampledFN):
utils.makeDir(resampledAudioPath)
sr = str(DEFAULT_FRAMERATE)
soxCmd = "%s %s -r %s %s rate -v 96k" % (soxPath, fn, sr,
resampledFN)
os.system(soxCmd)
if not os.path.exists(resampledFN):
raise IncompatibleSampleFrequencyError(framerate,
DEFAULT_FRAMERATE)
audiofile = wave.open(resampledFN, "r")
params = audiofile.getparams()
nchannels = params[0]
sampwidth = params[1]
framerate = params[2]
comptype = params[4]
compname = params[5]
# Extract the audio frames
audiofile.setpos(int(framerate * startT))
audioFrames = audiofile.readframes(int(framerate * (endT - startT)))
# Convert to single channel if needed
if singleChannelFlag is True and nchannels > 1:
audioFrames = audioop.tomono(audioFrames, sampwidth, 1, 0)
nchannels = 1
outParams = [nchannels, sampwidth, framerate, len(audioFrames),
comptype, compname]
outWave = wave.open(outputFN, "w")
outWave.setparams(outParams)
outWave.writeframes(audioFrames)
def read(self, buf, source_channels):
source_sample_width = pyaudio.get_sample_size(pyaudio.paInt16) * source_channels
audio = buf[3:]
try:
# sometimes the data received is incomplete so reusing state
# data from ratecv() sometimes results in errors
(audio, _) = audioop.ratecv(audio, source_sample_width, source_channels, 48000, self.listener.sample_rate, None)
audio = audioop.tomono(audio, self.listener.sample_width, 0.5, 0.5)
self.listener.read(audio)
except audioop.error, e:
logger.warn("Error preparing sample", exc_info=True)
def convert_wave_data(f_rate,frame_count,sample_width,channels,data):
""" Convert wave sample data into pleo format
"""
if channels==2: data = audioop.tomono(data,sample_width,1,1)
data = audioop.mul(data,sample_width,0.97999999999999998)
data = audioop.ratecv(data,sample_width,1,f_rate,11025,None,4,4)[0]
if sample_width==1:
data = audioop.bias(data,1,-128)
data = audioop.lin2lin(data,1,2)
data = audioop.mul(data,2,(1.0/256))
data = audioop.lin2adpcm(data,2,None)[0]
return (11025,frame_count,sample_width,1,data)
def localize():
#Returns an error between -1 and 1 that specifies which direction to move the head to
try:
#Start an infite loop that gets and analyzes audio data
alpha = .05 #The width of the sigmoid. Gotta tweak it a little.
#Average difference
avgDiff = 0
for i in range(0, 100):
l, data = audioInput.read()
if l > 0:
lchan = audioop.tomono(data, 2, 1, 0)
rchan = audioop.tomono(data, 2, 0, 1)
lmax = audioop.max(lchan, 2)
rmax = audioop.max(rchan, 2)
#lowpass filter
if lmax < 70 and rmax < 70:
continue
#calculage the difference in intensity. Positive difference means the source
#is further to the left
diff = lmax - rmax
avgDiff = avgDiff + diff
time.sleep(.001) #audio refresh rate
avgDiff = avgDiff / 1000
print avgDiff
#Logistic Sigmoid function!
z = math.exp(-1 * alpha * avgDiff)
error = (1 / (1 + z)) - 0.5
return error
except Exception, e:
print e
def calculate_max_min(self):
self.Sound.write(".temp.wav")
S = wave.open(".temp.wav","rb")
Length = S.getnframes()
BufferLen = Length / self.FragmentFactor
Stream = S.readframes(BufferLen)
while len(Stream) :
Stream = audioop.lin2lin(Stream,2,1)
Stream = audioop.tomono(Stream,1,1,-1)
min_val, max_val = audioop.minmax(Stream, 1)
self.Min_List.append(min_val)
self.Max_List.append(max_val)
Stream = S.readframes(BufferLen)
S.close()
os.remove(".temp.wav")