def makeIR(wav_in,wav_out,fs,duration,noise=0.025):
""" measures the response of a speaker (+amp+mic) and build an IR """
# step 1: full duplex playback and recording. Input: provided sweep wav file
# output: recorded time response
ecasound_cmd="ecasound -f:16,1,%i -a:1 -i jack,system,capture " + \
" -o /tmp/capture.wav -a:2 -i %s -o jack,system -t %i"
ecasound_cmd=ecasound_cmd%(int(fs),wav_in,int(duration))
# run capture
os.system(ecasound_cmd)
# load input and capture wave files
time.sleep(3)
f=wave.open(wav_in,'rb')
len1=f.getnframes()
#nc1=f.getnchannels()
#bp1=f.getsampwidth()
data=f.readframes(len1)
f.close()
Y1=scipy.float32(scipy.fromstring(data,dtype='int16'))
f=wave.open('/tmp/capture.wav','rb')
len2=f.getnframes()
#nc1=f.getnchannels()
#bp1=f.getsampwidth()
data=f.readframes(len2)
f.close()
Y2=scipy.float32(scipy.fromstring(data,dtype='int16'))
# truncate and normalize wave file
#(or we could pad the shortest to the longest... TODO!)
minlen = min([len1,len2])
Y2=Y2[0:minlen]
Y2=Y2/max(abs(Y2))
Y1=Y1[0:minlen]
Y1=Y1/max(abs(Y1))
# compute frequency response function as ration of both spectra
FRF=scipy.fft(Y2)/scipy.fft(Y1)
# compute impulse response as inverse FFT of FRF
IRraw=scipy.real(scipy.ifft(FRF))
# get rid of initial lag in IR
thr=max(abs(IRraw))*noise
offset=max([0 , min(min(scipy.where(abs(IRraw)>thr)))-5 ])
IR=IRraw[offset:-1]
IRnorm=IR/max(abs(IR))
# TODO: add post pro options such as low/high pass and decay
# write output IR
f = wave.open(wav_out, 'w')
f.setparams((1, 2, fs, 0, 'NONE', 'not compressed'))
maxVol=2**15-1.0 #maximum amplitude
wvData=""
for i in range(len(IRnorm)):
wvData+=pack('h', maxVol*IRnorm[i])
f.writeframes(wvData)
f.close()
def process(self, sbmp):
"""Core function"""
if sbmp == None:
# Oversize required
return self.size
# reverse height and width under advice
(ws, hs, ps) = shape = (sbmp.GetHeight(), sbmp.GetWidth(), 3)
simg = wx.ImageFromBitmap(sbmp)
sarray = scipy.array(scipy.fromstring(simg.GetData(), 'uint8'), self.dtype) / 255.0
sarray = scipy.rollaxis(scipy.reshape(sarray, shape), 2)
self.shape = sarray.shape
tarray = (self.withGPU if self.gpgpu else self.withCPU)(sarray)
mm = (sarray.min(), sarray.max(), tarray.min(), tarray.max())
#print '\t', sarray.shape, tarray.shape,
print type(sarray[0,0,0]), type(tarray[0,0,0]), mm
tarray = numpy.nan_to_num(tarray)
tarray /= max(tarray.max(), self.coefficient)
tarray = scipy.array((tarray * 255.0).tolist(), 'uint8')
tarray = scipy.dstack(tarray)
timg = wx.EmptyImage(ws, hs)
timg .SetData(tarray.tostring())
self.tbmp = timg.ConvertToBitmap()
return self.tbmp
def _read_nicolet(self, tmin, tmax):
"""Load Nicolet BMSI data."""
# print "_read_nicolet: tmin=", tmin, "tmax=", tmax
if tmin < 0:
tmin = 0
BYTES_PER_SAMPLE = self.channels * 2
indmin = int(self.freq * tmin)
NUMSAMPLES = os.path.getsize(self.fullpath) // BYTES_PER_SAMPLE
indmax = min(NUMSAMPLES, int(self.freq * tmax))
byte0 = indmin * BYTES_PER_SAMPLE
numbytes = (indmax - indmin) * BYTES_PER_SAMPLE
self.fh.seek(byte0)
data = fromstring(self.fh.read(numbytes), "h")
if sys.byteorder == "big":
data = data.byteswapped()
data = data.astype("d")
data.shape = -1, self.channels
if self.scale is not None:
data = self.scale * data
t = (1 / self.freq) * arange(indmin, indmax)
# print 'nic', data.shape
# print "_read_nicolet: t is " , t
return t, data
def readBlobs(self, stream):
length = len(stream)
blobs = []
while stream.tell() < length:
blobType = stream.readUInt32()
blobLength = stream.readUInt32()
if 0x0c == blobType:
zeros = stream.read(0x10)
if '\x00' * 0x10 != zeros:
raise Exception("Blob type 0xc decoding error")
supposedXor = stream.readUInt16()
twoZeros = stream.read(2)
if '\x00\x00' != twoZeros:
raise Exception("Blob type 0xc decoding error")
dataLength = stream.readUInt32()
zero = stream.readUInt32()
if 0 != zero:
raise Exception("Blob type 0xc decoding error")
somethingImportant = stream.readUInt32()
blobData = stream.read(dataLength)
dataArray = scipy.fromstring(blobData, scipy.uint16)
calcedXor = scipy.bitwise_xor.reduce(dataArray)
if calcedXor != supposedXor:
raise Exception("Worng XOR check in data blob %x != %x" % (calcedXor, supposedXor))
blobs.append((blobType, (blobData, somethingImportant)))
else:
blobData = stream.read(blobLength - 8)
blobs.append((blobType, blobData))
return blobs
def readData(self,ob):
f = open(ob.filename,'rb')
f.seek(int(self._header['STM image list']['Data offset']))
data = f.read(int(self._header['STM image list']['Data length']))
ob.d = scipy.fromstring(data,dtype=scipy.int16)
ob.d.shape = ob.XRes, ob.YRes
ob.d = scipy.flipud(ob.d)
def _read_float_array(self, fname):
"""Load an array of C floats."""
fh = file(fname, "rb")
data = fromstring(fh.read(), "f")
data.shape = -1, self.channels
return data
def encrypt( self, data, base=0x1000084 ):
if isinstance(data, (ObjectWithStreamBigEndian, ObjectWithStream)):
data = data.getRawData()
data = scipy.fromstring(data, scipy.uint16)
data.byteswap(True)
data = self.encryptChunk(data, base=base)
data.byteswap(True)
return data
def decrypt( self, data, base=0x1000084 ):
if isinstance(data, (ObjectWithStreamBigEndian, ObjectWithStream)):
data = data.getRawData()
data = scipy.fromstring(data, scipy.uint16)
data.byteswap(True)
data = self.decryptChunk(data, base)
data.byteswap(True)
data = scipy.bitwise_xor(data, data[-1])
return data.tostring()
def callback(in_data, frame_count, time_info, status):
# data = q.get()
d = self.wf.readframes(frame_count)
print frame_count
buf = scipy.fromstring(d, scipy.int16)
d2 = scipy.signal.lfilter(IIR_b, IIR_a, buf)
data = scipy.int16(d2).tostring()
return (data, pyaudio.paContinue)
def mainLoop(self):
global terminated, switch
# start Recording
audio = pyaudio.PyAudio()
stream = audio.open(format=FORMAT, channels=CHANNELS,
rate=FS, input=True,
frames_per_buffer=CHUNK)
# Cleaning the file. In case of a force quit, the file keep the last run data
open('out.bin', 'wb').close()
last_switch = 0
sound = []
while not terminated:
if recording:
# get audio samples
data = stream.read(CHUNK)
orig = fromstring(data, dtype="int16")
sound = np.append(sound, orig)
# Transform to frequency domain (FFT)
# reduce noise and transform back to time
originalfft = np.array(fft(orig))
fftSpec = abs(originalfft) / (CHUNK / 2)
fftSpec = fftSpec[:int(CHUNK / 2)]
xf = 1.0 * np.arange(0, FS / 2., FS / (1. * CHUNK))
# Spectrogram
self.img_array = np.roll(self.img_array, -1, 0)
self.img_array[-1:] = 10.0 * np.log10(fftSpec)
# Plotting Graphs
self.origWave.plot(orig, clear=True)
self.fftItem.plot(xf, fftSpec, clear=True)
self.specItem.setImage(self.img_array, autoLevels=False)
else:
if last_switch != switch:
last_switch = switch
frames = []
# Write in file
f = open('out.bin', 'ab')
for i in range(len(sound)):
f.write('%d ' % sound[i])
f.write('\n')
f.close()
sound = []
QtGui.QApplication.processEvents()
stream.stop_stream()
stream.close()
audio.terminate()
def readData(self,ob):
f = open(ob.filename,'rb')
f.seek(int(self._header['Data offset']))
points = int(self._header['SamplesT'])
datalength = ob.XRes * ob.YRes * points * 2
data = f.read(datalength)
ob.d = scipy.fromstring(data,dtype=scipy.int16)
ob.d.shape = ob.XRes, ob.YRes, points
ob.d = scipy.flipud(ob.d)
def readWaveAsFloat(wf):
length = wf.getnframes()
data = wf.readframes(length)
data = sp.fromstring(data, sp.int16)
# data = [1 if e is None else 0 for e in data]
# print(sum(data))
# data = np.append(data, [0] * 1024)
data = np.asarray(data, dtype="float64")
data = data / 32768.
print(data.shape)
return data
def read_signal(filename, winsize):
wf = wave.open(filename, 'rb')
n = wf.getnframes()
str = wf.readframes(n)
params = ((wf.getnchannels(), wf.getsampwidth(),
wf.getframerate(), wf.getnframes(),
wf.getcomptype(), wf.getcompname()))
siglen = ((int)(len(str) / 2 / winsize) + 1) * winsize
signal = sp.zeros(siglen, sp.int16)
signal[0:len(str) / 2] = sp.fromstring(str, sp.int16)
return [signal, params]
def open(self, wavefile):
self.wavefile = wave.open(wavefile, 'rb')
frames = self.wavefile.readframes(self.wavefile.getnframes())
# -1~+1の範囲に正規化
self.wavedata = sp.fromstring(frames, dtype='int16') / 32768.0
self.wavefile.rewind()
self.proc = subprocess.Popen(['aplay', '-q', wavefile])
# 再生開始時のずれ対策のため，start()まで待ち合わせ
self.proc.send_signal(signal.SIGSTOP)
self.start_time = time.time() # 再生開始時刻
self.stop_time = time.time() # 停止時刻
self.spend_time = 0.0 # 停止時間
def __init__(self, s):
self.data = fromstring(s[:18000], UInt8)
self.data.shape = -1, 18
self.timestamp = unpack('8s', s[18000:18008])[0].strip() # char [8]
self.rec_no = unpack('L', s[18008:18012])[0] # unsigned long
self.ux_time = unpack('L', s[18012:18016])[0] # unsigned long
self.smimage = unpack('8s', s[18016:18024])[0].strip() # char [8]
self.ox_rec_ptr = unpack('B', s[18024])[0] # unsigned char
self.oxes= unpack('64B', s[18025:18089]) # unsigned char [64]
self.rates = unpack('64B', s[18089:18153]) # unsigned char [64]
self.ox_acq_time = unpack('64H', s[18153:18281]) # unsigned int [64] (ushort?)
self.filstruct = unpack('150s', s[18281:18431])[0].strip() # char [150]
def read(fname, winsize):
if fname == "-":
wf = wave.open(sys.stdin, 'rb')
n = wf.getnframes()
str = wf.readframes(n)
params = ((wf.getnchannels(), wf.getsampwidth(),
wf.getframerate(), wf.getnframes(),
wf.getcomptype(), wf.getcompname()))
siglen = ((int)(len(str)/2/winsize) + 1) * winsize
signal = sp.zeros(siglen, sp.float32)
signal[0:len(str)/2] = sp.float32(sp.fromstring(str, sp.int16))/32767.0
return signal, params
else:
return read_signal(fname, winsize)
def load(filename, network=None):
r"""
Opens a 'csv' file, reads in the data, and adds it to the **Network**
Parameters
----------
filename : string (optional)
The name of the file containing the data to import. The formatting
of this file is outlined below.
Returns
-------
If no Network object is supplied then one will be created and returned.
"""
net = {}
with _read_file(filename=filename, ext='csv') as f:
a = _pd.read_table(filepath_or_buffer=f,
sep=',',
skipinitialspace=True,
index_col=False,
true_values=['T', 't', 'True', 'true',
'TRUE'],
false_values=['F', 'f', 'False', 'false',
'FALSE'])
# Now parse through all the other items
for item in a.keys():
element = item.split('.')[0]
prop = item.split('.', maxsplit=1)[1]
data = _sp.array(a[item].dropna())
if type(data[0]) is str:
N = _sp.shape(data)[0]
if '.' in data[0].split(' ')[0]: # Decimal means float
dtype = float
else:
dtype = int
temp = _sp.empty(_sp.shape(data), dtype=object)
for row in range(N):
temp[row] = _sp.fromstring(data[row], sep=' ', dtype=dtype)
data = _sp.vstack(temp)
else:
dtype = type(data[0])
net[element+'.'+prop] = data.astype(dtype)
if network is None:
network = OpenPNM.Network.GenericNetwork()
network = _update_network(network=network, net=net)
return network
def spectrumAnalyzer(self):
# FFT する信号の初期化
signal = zeros(fftLen, dtype=float)
# Update
print('音声入力ループ')
sound_data = []
# 適当に5秒間実行して終了
for n in xrange(0, self.fs * 7 / self.chunk):
try:
# dataは文字列型
data = self.stream.read(self.chunk)
except IOError as ex:
# よくわからんけど、しばらく実行していると結構な頻度でミスってる
if ex[1] != pyaudio.paInputOverflowed:
raise
data = '\x00' * self.chunk
print('errorrrrrrrrrrrrrrrrrr')
num_data = fromstring(data, dtype='int16')
signal = roll(signal, - self.chunk)
signal[- len(num_data):] = num_data
fftspec = my_fft(signal)
spec = abs(fftspec[1: fftLen / 2 + 1]) * signal_scale # スペクトル
# print('Max : %.5f' % freq_list[np.argmax(spec)])
if max(spec) >= 2000 and np.argmax(spec) > 3:
print('spec: %s' % str(max(spec)))
max_list_num = np.argmax(spec)
print('Max : %8.3f , %s, %s' % (
self.freq_list[max_list_num][0],
self.my_piano.octa_mark[self.freq_list[max_list_num][1]],
self.my_piano.onkai[self.freq_list[max_list_num][2]])
)
sound_data.append(phone_scale[self.freq_list[max_list_num][2]])
else:
print('none')
sound_data.append('N')
self.specItem.plot(spec, clear=True)
QtGui.QApplication.processEvents()
self.save_score_data(sound_data, self.chunk/float(self.fs), OUTPUT_FILE_NAME)
self.stream.close()
self.p.terminate()
def img_from_fig(fig):
"""produce :Image: instance from :fig:
:type fig: matplotlib.figure.Figure
:param fig: input figure
:rtype: Image.Image
"""
if not isinstance(fig, Figure):
raise TypeError('fig must be a %s' % Figure)
if fig.canvas is None:
cvs = FigureCanvasAgg(fig)
fig.canvas.draw()
rgb = fig.canvas.tostring_rgb()
rgb = sp.fromstring(rgb, dtype=sp.uint8)
rgb.shape = map(int, fig.bbox.bounds[2:]) + [3]
return MRPlot.img_from_rgb(rgb)
def writeBlobs(self, outputStream, blobs, address, plain):
for blobType, blobData in blobs:
outputStream.writeUInt32(blobType)
if 0x0c == blobType:
codeData, somethingImportant = blobData
outputStream.writeUInt32(len(codeData) + 0x20 + 8)
outputStream.write('\x00' * 0x10)
dataArray = scipy.fromstring(codeData, scipy.uint16)
calcedXor = scipy.bitwise_xor.reduce(dataArray)
outputStream.writeUInt16(calcedXor)
outputStream.writeUInt16(0)
outputStream.writeUInt32(len(codeData))
outputStream.writeUInt32(0)
outputStream.writeUInt32(somethingImportant)
outputStream.write(codeData)
else:
outputStream.writeUInt32(len(blobData) + 8)
outputStream.write(blobData)
def from_data(data):
"""produce a SimPkg from bytedata
:Parameters:
data : str
The data to produce the package from.
"""
# length check
if len(data) < SimPkg.HLEN:
raise ValueError('length < SimPkg.HLEN')
# read header
idx = SimPkg.HLEN
tid, ident, frame, nitems = unpack(SimPkg.HDEF, data[:idx])
cont = []
# content loop
while idx < len(data):
# read contents header
dim0, dim1, nbytes, dtype_str = unpack(ContentItem.HDEF, data[idx:idx + ContentItem.HLEN])
idx += ContentItem.HLEN
# read content data
cont_item = N.fromstring(
data[idx:idx + nbytes],
dtype=N.dtype(dtype_str)
)
if dim0 >= 0:
if dim1 >= 0:
dim = [dim0, dim1]
else:
dim = [dim0]
cont_item.shape = dim
cont.append(cont_item)
idx += nbytes
# return
assert len(cont) == nitems, 'cont list length (%s) does not match nitems (%s)!' % (len(cont), nitems)
return SimPkg(tid, ident, frame, tuple(cont))
def _loadEGM96():
"""load the EGM96 geoid model into a spline object"""
# load the data resource file into a string
flc = resource_string(__name__, "data/egm96.dac")
# setup basic coordinates
lon = sp.linspace(0, 2 * sp.pi, 1440, False)
lat = sp.linspace(0, sp.pi, 721)
# parse the raw data string
data = sp.fromstring(flc, sp.dtype(sp.int16).newbyteorder("B"),
1038240).reshape((lat.size, lon.size)) / 100.0
# interpolate data
lut = RectSphereBivariateSpline(lat[1: -1], lon, data[1: -1],
pole_values=(sp.mean(data[1]), sp.mean(data[-1])))
return lut
def threader():
for i in range(100):
# 時間計測
start_b = time.time()
# オーディオデータの呼び出し
d = self.wf.readframes(BUFFER_SIZE) # str
# strをintに変換
buf = scipy.fromstring(d, scipy.int16)
# フィルタリング
data = scipy.signal.lfilter(IIR_b, IIR_a, buf)
# strに変換
self.buffer = scipy.int16(data).tostring()
q.put(self.buffer)
# パフォーマンス
elapsed_time = time.time() - start_b
print("buffer_time:{0}".format(elapsed_time))
def _initialize_file(self, filename, **kwargs):
# open file
self.fp = open(filename, 'r')
# read header dbconfig
self.header = _ATF_H(self.fp)
self.nchan = len(self.header.signals_exported)
self.ndata = self.header.datasets[1] - 1 / self.nchan
# read data
data = sp.fromstring(self.fp.read(), dtype=self.dtype, sep='\t')
data.shape = (
data.shape[0] / self.header.datasets[1],
self.header.datasets[1]
)
data = data.T
self._sample_times = data[0, :]
self._data = data[1:, :]
del data
def cut_wav(filename, time):
# timeの単位は[sec]
# ファイルを読み出し
#wavf = filename + '.wav'
wavf = filename
wr = wave.open(wavf, 'r')
# waveファイルが持つ性質を取得
ch = wr.getnchannels()
width = wr.getsampwidth()
fr = wr.getframerate()
fn = wr.getnframes()
total_time = 1.0 * fn / fr
integer = math.floor(total_time) # 小数点以下切り捨て
t = int(time) # 秒数[sec]
frames = int(ch * fr * t)
num_cut = int(integer // t)
data = wr.readframes(wr.getnframes())
wr.close()
X = fromstring(data, dtype=int16)
for i in range(num_cut):
# print(i)
# 出力データを生成
outf = '001_bunkatsu/' + str(i) + '.wav'
start_cut = i * frames
end_cut = i * frames + frames
# print(start_cut)
# print(end_cut)
Y = X[start_cut:end_cut]
outd = struct.pack("h" * len(Y), *Y)
# 書き出し
ww = wave.open(outf, 'w')
ww.setnchannels(ch)
ww.setsampwidth(width)
ww.setframerate(fr)
ww.writeframes(outd)
ww.close()
return num_cut
def importWave(self):
"""Wave file to ndarray"""
wf = wave.open(self.filename, 'rb')
waveframes = wf.readframes(wf.getnframes())
self.framerate = wf.getframerate()
data = sp.fromstring(waveframes, sp.int16)
self.duration = float(wf.getnframes()) / self.framerate
if(wf.getnchannels() == 2):
left = sp.array([data[i] for i in range(0, data.size, 2)])
right = sp.array([data[i] for i in range(1, data.size, 2)])
left = sp.int32(left); right = sp.int32(right)
data = sp.int16(left+right) / 2
if(self.fs == None):
self.fs = self.framerate
else:
#data = self.resample(data, data.size*(self.fs/self.framerate))
data = ssig.decimate(data, int(self.framerate/self.fs))
self.duration_list = sp.arange(0, self.duration, 1./self.fs)
data = ssig.detrend(data)
return data
def get_data(sound):
buffer = sound.readframes(sound.getnframes())
buffer = fromstring(buffer, dtype=int16)
data_a = []
data_b = []
channels = sound.getnchannels()
print(len(buffer))
for i in buffer:
tmp = 32768 + i
a1 = tmp // 256
a2 = a1
a1 = a1 // 16
a2 = a2 % 16
b1 = tmp % 256
b2 = b1
b1 = b1 // 16
b2 = b2 % 16
data_a.append(a1 + b1 * 16)
data_b.append(a2 + b2 * 16)
data = [data_a, data_b, channels]
return data
def __init__(self, fp):
"""
:type fp: file
:param fp: open file at seek(0)
"""
# version
self.version = fp.readline().strip('\'\"\r\n').split()
if self.version != ['ATF', '1.0']:
raise DataFileError('wrong version: %s' % self.version)
# data set structure
self.datasets = fp.readline().strip('\'\"\r\n').split()
self.datasets = map(int, self.datasets)
if len(self.datasets) != 2:
raise DataFileError('invalid file structure: %s' %
str(self.datasets))
self.signals_exported = None
self.sweep_times = None
self.dbconfig = {}
# signal names
for _ in xrange(self.datasets[0]):
line = fp.readline().strip('\'\"\r\n')
if line.startswith('SignalsExported'):
self.signals_exported = line.split('=')[-1].split(',')
elif line.startswith('SweepStartTimesMS'):
self.sweep_times = sp.fromstring(line.split('=')[1], sep=',')
else:
# TODO: if we need other header infos, read in here
pass
if self.signals_exported is None:
raise DataFileError('could not get signal count and names!')
# column headers
self.col_headers = fp.readline().strip('\r\n').split('\t')[1:]
self.col_headers =\
map(str.strip, self.col_headers, ['\'\"'] * len(self.col_headers))
def get_data(sound):
buffer = sound.readframes(sound.getnframes())
buffer = fromstring(buffer, dtype=int16)
data_a = []
data_b = []
channels = sound.getnchannels()
print(len(buffer))
for i in buffer:
a_0b = ""
b_0b = ""
tmp = 32768 + i
for j in range(16):
if j % 2 == 0:
a_0b += str(tmp & 0b1)
tmp >>= 1
else:
b_0b += str(tmp & 0b1)
tmp >>= 1
data_a.append(int(a_0b, 2))
data_b.append(int(b_0b, 2))
data = [data_a, data_b, channels]
return data
def _loadEGM96():
"""load the EGM96 geoid model into a spline object"""
#load the data resource file into a string
flc = resource_string(__name__, "data/egm96.dac")
#setup basic coordinates
lon = sp.linspace(0, 2 * sp.pi, 1440, False)
lat = sp.linspace(0, sp.pi, 721)
#parse the raw data string
data = sp.fromstring(flc,
sp.dtype(sp.int16).newbyteorder("B"),
1038240).reshape((lat.size, lon.size)) / 100.0
#interpolate the bad boy
lut = RectSphereBivariateSpline(lat[1:-1],
lon,
data[1:-1],
pole_values=(sp.mean(data[1]),
sp.mean(data[-1])))
return lut
def WavRead(FNAME, dtype=np.float64):
'''
Load WAVE Format file
<<Input>>
FNAME ... File Name
dtype ... Data type
<<Output>>
x ... Waveform
'''
### File Open ###
try:
wf = wave.open(FNAME, 'rb')
except:
print "FILE I/O error!"
sys.exit()
### Load all data ###
data = wf.readframes(wf.getnframes())
x = sp.fromstring(data, sp.int16)
wf.close()
return x
def load(cls, filename, network=None):
r"""
"""
net = {}
# ---------------------------------------------------------------------
# Parse the link1 file
filename = cls._parse_filename(filename=filename, ext='am')
with open(filename, mode='r') as f:
Np = None
Nt = None
while (Np is None) or (Nt is None):
s = f.readline()[:-1].split(' ')
if s[0] == 'define':
if s[1] == 'VERTEX':
Np = int(s[2])
if s[1] == 'EDGE':
Nt = int(s[2])
net = {}
propmap = {}
typemap = {}
shapemap = {}
while True:
s = f.readline()[:-1].split(' ')
if s[0] == 'VERTEX':
dshape = [Np]
if s[2].endswith(']'):
ncols = int(s[2].split('[', 1)[1].split(']')[0])
dshape.append(ncols)
dtype = s[2].split('[')[0]
temp = sp.zeros(dshape, dtype=dtype)
net['pore.'+s[3]] = temp
key = int(s[-1].replace('@', ''))
propmap[key] = 'pore.'+s[3]
typemap[key] = dtype
shapemap[key] = dshape
elif s[0] == 'EDGE':
dshape = [Nt]
if s[2].endswith(']'):
ncols = int(s[2].split('[', 1)[1].split(']')[0])
dshape.append(ncols)
dtype = s[2].split('[')[0]
temp = sp.zeros(dshape, dtype=dtype)
net['throat.'+s[3]] = temp
key = int(s[-1].replace('@', ''))
propmap[key] = 'throat.'+s[3]
typemap[key] = dtype
shapemap[key] = dshape
elif s[0] == '#':
break
s = f.read().split('@')
for key in propmap.keys():
if key in s:
data = s[key].split('\n')[1:]
data = ' '.join(data)
arr = sp.fromstring(data, dtype=typemap[key], sep=' ')
arr = sp.reshape(arr, newshape=shapemap[key])
net[propmap[key]] = arr
# End file parsing
net['pore.coords'] = net['pore.VertexCoordinates']
net['throat.conns'] = sp.sort(net['throat.EdgeConnectivity'], axis=1)
if network is None:
network = GenericNetwork()
network = cls._update_network(network=network, net=net)
return network.project
import pyaudio
from scipy import fromstring, int16
from matplotlib import pyplot as pl
CHUNK=1024
p=pyaudio.PyAudio()
input_device_index=0
stream=p.open(format=pyaudio.paInt16,
channels=1,
rate=44100,
frames_per_buffer=CHUNK,
input=True)
i=0
while stream.is_active():
try:
input=stream.read(CHUNK)
num_data=fromstring(input, dtype="int16")/32768.0
#print(num_data)
#print('最大値:'+str(num_data.max()))
if(num_data.max()>=0.25):
print('BIG!!!!!!!!'+str(i))
i+=1
#pl.plot(num_data)
#pl.draw()
#pl.pause(0.01)
#pl.cla()
except KeyboardInterrupt:
#pl.close()
break
def exec_wav_spectrum():
# 仮想COMポートなのでボーレートは無意味
ser = serial.Serial('COM52', 1000000)
hamming_win = sp.hamming(FFT_SIZE)
wavefile = wave.open('mtank.wav', 'rb')
frames = wavefile.readframes(wavefile.getnframes())
# ±1の範囲に正規化
wavedata = sp.fromstring(frames, dtype='int16') / 32768.0
wavefile.rewind()
pa = pyaudio.PyAudio()
print('WAV: [OK]')
count = 0
gain = 1.00
fps = 60.0
def callback(in_data, frame_count, time_info, status):
frames = wavefile.readframes(frame_count)
wave = sp.fromstring(frames, dtype='int16')
# 再生音量にもゲイン適用
gained_wave = wave * gain
gained_wave = np.clip(gained_wave, -32768, +32767)
data = bytes(gained_wave.astype(np.int16))
return (data, pyaudio.paContinue)
stream = pa.open(format = pa.get_format_from_width(wavefile.getsampwidth()),
channels = wavefile.getnchannels(),
rate = wavefile.getframerate(),
output = True,
stream_callback = callback)
stream.start_stream()
start_time = time.time()
while (stream.is_active()):
"""
モード切り替え判定
"""
ser.write('md\n'.encode())
line = ser.readline()
if (int(line) is 1):
print('Mode Switch!')
break
# ゲイン調整
if ((count % 10) == 0):
gain = read_gain(ser)
print('gain=%.2f' % gain)
# 経過時間からフレーム箇所を特定してそこをFFTする
# data[1024](要正規化) --> spec[24]
frame_time = time.time() - start_time
frame_pos = int(frame_time * wavefile.getframerate())
fft_input = wavedata[frame_pos : frame_pos + FFT_SIZE]
# ゲイン適用
fft_input = fft_input * gain
if (len(fft_input) < FFT_SIZE):
fft_input = np.zeros(FFT_SIZE)
fft_output = sp.fft(fft_input * hamming_win)
########################################
y = []
toStep = 0
fromStep = 0
specSize = 513
for i in range(len(bandHz)):
bandStep = bandHz[i]
toStep += bandStep
if (toStep > specSize):
toStep = specSize
bandAve = 0.0
j = fromStep
while (j < toStep):
bandDB = 0.0
if (abs(fft_output[j]) >= 0.001):
bandDB = 2 * (20 * ((math.log10(abs(fft_output[j])))))
bandDB = (20 * ((math.log10(abs(fft_output[j])))))
if (bandDB < 0):
bandDB = 0
bandAve += bandDB
j += 1
# 平均値
bandAve /= bandStep
fromStep = toStep
# 最終加工
bandAve /= 1.5
y.append(int(bandAve))
########################################
spec = y
panel = spec_to_panel(spec)
# 更新回数表示
panel[0][23] = num_to_pattern[count // 1 % 10]
panel[0][22] = num_to_pattern[count // 10 % 10]
panel[0][21] = num_to_pattern[count // 100 % 10]
panel[0][20] = num_to_pattern[count // 1000 % 10]
xfer_data = panel_to_command(panel, 0x01)
write_display(ser, xfer_data)
# fpsの値に合わせて規定時間になるまで待つ
expected_time = start_time + (((1000.0 / fps) * count) / 1000)
while (time.time() < expected_time):
time.sleep(0.001)
print('_', end='')
count += 1
elapsed_time = time.time() - start_time
ser.close()
stream.stop_stream()
stream.close()
wavefile.close()
pa.terminate()
if (elapsed_time > 0):
print('%f [s]' % elapsed_time)
print('fps = %f' % (count / elapsed_time))
print('Done.')
def train_loop(model, optimizer, train_set, scheduler=None):
num_mb = len(train_set) // hp.batch_size
if scheduler:
scheduler.step(epoch)
for i in range(num_mb):
# input lmfb (B x T x (F x frame_stacking))
xs = []
# target symbols
ts = []
# onehot vector of target symbols (B x L x NUM_CLASSES)
ts_onehot = []
# vector of target symbols for label smoothing (B x L x NUM_CLASSES)
ts_onehot_LS = []
# input lengths
emo = []
emo_onehot = []
emo_onehot_LS = []
lengths = []
ts_lengths = []
temp = []
temp_length = []
for j in range(hp.batch_size):
s = train_set[i * hp.batch_size + j].strip()
if hp.ASR:
x_file, laborg = s.split(' ', 1)
elif hp.dist:
x_file, laborg, labemo, labdist = s.split('\t')
laborg = laborg.strip()
labemo = labemo.strip()
labdist = labdist.strip()
else:
x_file, laborg, labemo = s.split('\t')
laborg = laborg.strip()
labemo = labemo.strip()
#if len(laborg) == 0:
# laborg = "2 0 1"
if '.htk' in x_file:
#mean = np.load("/n/work1/feng/src/htk/mean.npy")
#var = np.load("/n/work1/feng/src/htk/var.npy")
cpudat = load_dat(x_file)
cpudat = cpudat[:, :hp.lmfb_dim]
#cpudat = (cpudat-mean)/var
#print(mean)
elif '.npy' in x_file:
#mean = np.load("/n/work1/feng/data/swb/mean.npy")
#var = np.load("/n/work1/feng/data/swb/var.npy")
cpudat = np.load(x_file)
#cpudat = (cpudat-mean)/var
elif '.wav' in x_file:
with wave.open(x_file) as wf:
dat = wf.readframes(wf.getnframes())
y = fromstring(dat, dtype=int16)[:, np.newaxis]
y_float = y.astype(np.float32)
cpudat = (y_float - np.mean(y_float)) / np.std(y_float)
tmp = copy.deepcopy(cpudat)
print("{} {}".format(x_file, cpudat.shape[0]))
if hp.frame_stacking > 1 and hp.encoder_type != 'Wave':
cpudat, newlen = frame_stacking(cpudat, hp.frame_stacking)
newlen = cpudat.shape[0]
if hp.encoder_type == 'CNN':
cpudat_split = np.split(cpudat, 3, axis=1)
cpudat = np.hstack((cpudat_split[0].reshape(newlen, 1, 80),
cpudat_split[1].reshape(newlen, 1, 80),
cpudat_split[2].reshape(newlen, 1, 80)))
newlen = cpudat.shape[0]
lengths.append(newlen)
xs.append(cpudat)
temp.append(tmp)
temp_length.append(tmp.shape[0])
cpulab = np.array([int(i) for i in laborg.split(' ')],
dtype=np.int32)
#print(cpulab)
cpulab_onehot = onehot(cpulab, hp.num_classes)
ts.append(cpulab)
ts_lengths.append(len(cpulab))
ts_onehot.append(cpulab_onehot)
ts_onehot_LS.append(0.9 * cpulab_onehot +
0.1 * 1.0 / hp.num_classes)
if hp.dist and hp.ASR == False:
cpuemo = np.array([int(x) for x in labemo], dtype=np.int32)
emotion_onehot = onehot_dist(labdist, hp.num_emotion)
emo_onehot.append(emotion_onehot)
emo_onehot_LS.append(0.9 * emotion_onehot +
0.1 * 1.0 / hp.num_emotion)
emo.append(cpuemo)
elif hp.ASR == False:
cpuemo = np.array([int(x) for x in labemo], dtype=np.int32)
emotion_onehot = onehot(cpuemo, hp.num_emotion)
emo_onehot.append(emotion_onehot)
emo_onehot_LS.append(0.9 * emotion_onehot +
0.1 * 1.0 / hp.num_emotion)
emo.append(cpuemo)
if hp.baseline_type != 'lim_BLSTM':
temp, temp_length = xs, lengths
if hp.ASR:
xs, lengths, ts, ts_onehot, ts_onehot_LS, ts_lengths = sort_pad(
hp.batch_size, xs, lengths, ts, ts_onehot, ts_onehot_LS,
ts_lengths)
youtput_in_Variable = model(xs, lengths, ts_onehot, [], [])
loss = 0.0
if hp.decoder_type == 'Attention':
for k in range(hp.batch_size):
num_labels = ts_lengths[k]
loss += label_smoothing_loss(
youtput_in_Variable[k][:num_labels],
ts_onehot_LS[k][:num_labels], 1) / num_labels
print('loss = {}'.format(loss.item()))
elif hp.baseline:
xs, lengths, ts, ts_onehot, ts_onehot_LS, ts_lengths, emo, emo_onehot, emo_onehot_LS, temp = sort_pad(
hp.batch_size, xs, lengths, ts, ts_onehot, ts_onehot_LS,
ts_lengths, emo, emo_onehot, emo_onehot_LS, temp, temp_length)
if hp.baseline_type == 'CNN_BLSTM' or hp.baseline_type == 'lim_BLSTM':
onehot_length = temp.size(2)
xs_new = torch.zeros((hp.batch_size, 750, onehot_length))
for i in range(hp.batch_size):
feature_length = temp.size(1)
if feature_length > 750:
xs_new.data[:, :750, :] = temp.data[:, :750, :]
else:
xs_new.data[:, :
feature_length, :] = temp.data[:, :
feature_length, :]
emotion_in_Variable = model(xs_new.to(DEVICE), [])
else:
#youtput_in_Variable, emotion_in_Variable = model(xs, lengths, ts_onehot, emo_onehot, [])
emotion_in_Variable = model(xs, lengths)
loss = 0.0
if hp.decoder_type == 'Attention':
#print(emo)
#print(emotion_in_Variable[:,:hp.num_emotion])
loss += F.cross_entropy(
emotion_in_Variable[:, :hp.num_emotion], emo.to(DEVICE))
#for k in range(hp.batch_size):
#num_labels = ts_lengths[k]
#loss += label_smoothing_loss(youtput_in_Variable[k][:num_labels], ts_onehot_LS[k][:num_labels],1) / num_labels
#print(emotion_in_Variable[k][:hp.num_emotion])
#loss += F.cross_entropy(emotion_in_Variable[k][:hp.num_emotion], emo)
print('loss = {}'.format(loss.item()))
elif hp.text_based:
xs, lengths, ts, ts_onehot, ts_onehot_LS, ts_lengths, emo, emo_onehot, emo_onehot_LS, temp = sort_pad(
hp.batch_size, xs, lengths, ts, ts_onehot, ts_onehot_LS,
ts_lengths, emo, emo_onehot, emo_onehot_LS, temp, temp_length)
emotion_in_Variable = model(ts.to(DEVICE), ts_lengths.to(DEVICE))
loss = 0.0
if hp.decoder_type == 'Attention':
#print(emo)
#print(emotion_in_Variable[:,:hp.num_emotion])
loss += F.cross_entropy(
emotion_in_Variable[:, :hp.num_emotion], emo.to(DEVICE))
#for k in range(hp.batch_size):
#num_labels = ts_lengths[k]
#loss += label_smoothing_loss(youtput_in_Variable[k][:num_labels], ts_onehot_LS[k][:num_labels],1) / num_labels
#print(emotion_in_Variable[k][:hp.num_emotion])
#loss += F.cross_entropy(emotion_in_Variable[k][:hp.num_emotion], emo)
print('loss = {}'.format(loss.item()))
elif hp.combined:
#seq1 = []
#seq2 = []
#seq1, seq2, xs, lengths, ts, ts_onehot, ts_onehot_LS, ts_lengths, emo, emo_onehot, emo_onehot_LS, \
#xs1, lengths1, ts1, ts_onehot1, ts_onehot_LS1, ts_lengths1, emo1, emo_onehot1, emo_onehot_LS1 \
#= sort_pad(hp.batch_size, xs, lengths, ts, ts_onehot, ts_onehot_LS, ts_lengths, emo, emo_onehot, emo_onehot_LS)
xs, lengths, ts, ts_onehot, ts_onehot_LS, ts_lengths, emo, emo_onehot, emo_onehot_LS, temp = sort_pad(
hp.batch_size, xs, lengths, ts, ts_onehot, ts_onehot_LS,
ts_lengths, emo, emo_onehot, emo_onehot_LS, temp, temp_length)
if hp.baseline_type == 'CNN_BLSTM' or hp.baseline_type == 'lim_BLSTM':
onehot_length = temp.size(2)
xs_new = torch.zeros((hp.batch_size, 750, onehot_length))
for i in range(hp.batch_size):
feature_length = temp.size(1)
if feature_length > 750:
xs_new.data[:, :750, :] = temp.data[:, :750, :]
else:
xs_new.data[:, :
feature_length, :] = temp.data[:, :
feature_length, :]
emotion_in_Variable = model(xs_new.to(DEVICE),
[], ts.to(DEVICE),
ts_lengths.to(DEVICE))
else:
emotion_in_Variable = model(xs.to(DEVICE), lengths,
ts.to(DEVICE),
ts_lengths.to(DEVICE))
loss = 0.0
if hp.decoder_type == 'Attention':
#print(emo)
#print(emotion_in_Variable[:,:hp.num_emotion])
#for i in range(hp.batch_size):
# for j in range(hp.batch_size):
# if seq1[j] == i:
# break
# temp = emo[i]
# emo[i] = emo[j]
# emo[j] = temp
loss += F.cross_entropy(
emotion_in_Variable[:, :hp.num_emotion], emo.to(DEVICE))
#for k in range(hp.batch_size):
#num_labels = ts_lengths[k]
#loss += label_smoothing_loss(youtput_in_Variable[k][:num_labels], ts_onehot_LS[k][:num_labels],1) / num_labels
#print(emotion_in_Variable[k][:hp.num_emotion])
#loss += F.cross_entropy(emotion_in_Variable[k][:hp.num_emotion], emo)
print('loss = {}'.format(loss.item()))
elif hp.combined_ASR or hp.ASR_based:
xs, lengths, ts, ts_onehot, ts_onehot_LS, ts_lengths, emo, emo_onehot, emo_onehot_LS, temp = sort_pad(
hp.batch_size, xs, lengths, ts, ts_onehot, ts_onehot_LS,
ts_lengths, emo, emo_onehot, emo_onehot_LS, temp, temp_length)
if hp.baseline_type == 'CNN_BLSTM' or hp.baseline_type == 'lim_BLSTM':
onehot_length = temp.size(2)
xs_new = torch.zeros((hp.batch_size, 750, onehot_length))
for i in range(hp.batch_size):
feature_length = temp.size(1)
if feature_length > 750:
xs_new.data[:, :750, :] = temp.data[:, :750, :]
else:
xs_new.data[:, :
feature_length, :] = temp.data[:, :
feature_length, :]
youtput_in_Variable, emotion_in_Variable = model(
xs, lengths, ts_onehot, emo_onehot, xs_new.to(DEVICE))
else:
youtput_in_Variable, emotion_in_Variable = model(
xs, lengths, ts_onehot, emo_onehot, [])
loss = 0.0
if hp.decoder_type == 'Attention':
#print(emo)
#print(emotion_in_Variable[:,:hp.num_emotion])
loss += F.cross_entropy(
emotion_in_Variable[:, :hp.num_emotion],
emo.to(DEVICE)) * 0.8
print(loss)
for k in range(hp.batch_size):
num_labels = ts_lengths[k]
loss += label_smoothing_loss(
youtput_in_Variable[k][:num_labels],
ts_onehot_LS[k][:num_labels], 1) / num_labels * 0.2
#print(emotion_in_Variable[k][:hp.num_emotion])
#loss += F.cross_entropy(emotion_in_Variable[k][:hp.num_emotion], emo)
print('loss = {}'.format(loss.item()))
sys.stdout.flush()
optimizer.zero_grad()
# backward
loss.backward()
clip = 1.0
torch.nn.utils.clip_grad_value_(model.parameters(), clip)
# optimizer update
optimizer.step()
loss.detach()
torch.cuda.empty_cache()
synthfile = 'tools/sound/noisy.wav'
synth = wave.open(synthfile, 'wb')
synth.setnchannels(1)
synth.setsampwidth(2)
synth.setframerate(samplingrate)
remain = sound.getnframes()
while remain > 0:
s = min(chunk, remain)
#read frames
data_sound = sound.readframes(s)
data_noise = noise.readframes(s)
#convert
ary_sound = sp.fromstring(data_sound, sp.int16)
ary_noise = sp.fromstring(data_noise, sp.int16)
int32_ary_sound = sp.int32(ary_sound)
int32_ary_noise = sp.int32(ary_noise)
ary2 = sp.int16(int32_ary_sound + int32_ary_noise)
data2 = ary2.tostring()
synth.writeframes(data2)
remain = remain - s
sound.close()
noise.close()
synth.close()
infile = 'tools/sound/noisy.wav'
signal, params = read_signal(infile, WINSIZE)
nf = len(signal) / (WINSIZE / 2) - 1
synthfile = 'tools/sound/noisy.wav'
synth = wave.open(synthfile, 'wb')
synth.setnchannels(1)
synth.setsampwidth(2)
synth.setframerate(samplingrate)
remain = sound.getnframes()
while remain > 0:
s = min(chunk, remain)
# read frames
data_sound = sound.readframes(s)
data_noise = noise.readframes(s)
# convert
ary_sound = sp.fromstring(data_sound, sp.int16)
ary_noise = sp.fromstring(data_noise, sp.int16)
int32_ary_sound = sp.int32(ary_sound)
int32_ary_noise = sp.int32(ary_noise)
ary2 = sp.int16(int32_ary_sound + int32_ary_noise)
data2 = ary2.tostring()
synth.writeframes(data2)
remain = remain - s
sound.close()
noise.close()
synth.close()
infile = 'tools/sound/noisy.wav'
signal, params = read_signal(infile, WINSIZE)
nf = len(signal) / (WINSIZE / 2) - 1
p = pyaudio.PyAudio()
filename = "121_dr_bpm080_4-4_rock.wav"
wf = wave.open(filename, "rb")
stream = p.open(format=pyaudio.paInt16,
channels=1,
rate=44100,
frames_per_buffer=CHUNK,
input=True,
output=True)
#リアルタイム録音再生
while stream.is_active():
try:
input = stream.read(CHUNK)
num_data = fromstring(input, dtype=int16)
print(num_data)
list = np.array(num_data)
output = stream.write(list)
pyplot.plot(list)
pyplot.draw()
pyplot.pause(0.05)
pyplot.cla()
except KeyboardInterrupt:
pyplot.close()
break
if (wf.getnchannels() == 2):
left = num_data[::2]
right = num_data[1::2]
def mainLoop():
global terminated, fftSpec
# start Recording
audio = pyaudio.PyAudio()
stream = audio.open(format=FORMAT, channels=CHANNELS,
rate=FS, input=True,
frames_per_buffer=CHUNK)
### Application Creation
### Main window
mainWindow = myMainWindow()
mainWindow.show()
mainWindow.setWindowTitle("Spectrum Analyzer") # Title
mainWindow.resize(1300, 500) # Size
### Campus
centralWid = QtGui.QWidget()
mainWindow.setCentralWidget(centralWid)
layH = QtGui.QHBoxLayout()
centralWid.setLayout(layH)
LeftWidget = QtGui.QWidget()
RightWidget = QtGui.QWidget()
layH.addWidget(LeftWidget)
layH.addWidget(RightWidget)
LeftlayV = QtGui.QVBoxLayout()
LeftWidget.setLayout(LeftlayV)
RightlayV = QtGui.QVBoxLayout()
RightWidget.setLayout(RightlayV)
### Original Wave display widget
waveWid = pg.PlotWidget(title="Original Wave")
origWave = waveWid.getPlotItem()
origWave.setMouseEnabled(y=False) # to not be moved to the y-axis direction
origWave.setYRange(-10000, 10000)
origWave.setXRange(0, 512, padding=0)
### Axis
specAxis = origWave.getAxis("bottom")
specAxis.setLabel("Samples")
LeftlayV.addWidget(waveWid)
### Spectrum display widget
fftWid = pg.PlotWidget(title="FFT")
fftItem = fftWid.getPlotItem()
fftItem.setMouseEnabled(y=False) # to not be moved to the y-axis direction
fftItem.setYRange(0, 3000)
fftItem.setXRange(0, FS/2, padding=0)
### Axis
specAxis = fftItem.getAxis("bottom")
specAxis.setLabel("Frequency [Hz]")
LeftlayV.addWidget(fftWid)
### Spectogram
specWid = pg.PlotWidget()
specItem = pg.ImageItem()
specWid.addItem(specItem)
img_array = np.zeros((100, CHUNK // 2))
# bipolar colormap
pos = np.array([0., 1., 0.5, 0.25, 0.75])
color = np.array([[0, 255, 255, 255], [255, 255, 0, 255], [0, 0, 0, 255], (0, 0, 255, 255), (255, 0, 0, 255)],
dtype=np.ubyte)
cmap = pg.ColorMap(pos, color)
lut = cmap.getLookupTable(0.0, 1.0, 256)
# set colormap
specItem.setLookupTable(lut)
specItem.setLevels([-50, 40])
# setup the correct scaling for y-axis
freq = np.arange((CHUNK / 2) + 1) / (float(CHUNK) / FS)
yscale = 1.0 / (img_array.shape[1] / freq[-1])
specItem.scale((1. / FS) * CHUNK, yscale)
specWid.setLabel('left', 'Frequency', units='Hz')
RightlayV.addWidget(specWid)
### Window display
mainWindow.show()
frames = []
open('recorded.bin', 'w').close()
file = open('recorded.bin', 'a')
while not terminated:
orig = np.array([])
# get audio samples
data = stream.read(CHUNK)
frames.append(data)
orig = fromstring(data, dtype="int16")
# Transform to frequency domain (FFT)
originalfft = np.array(fft(orig))
fftSpec = abs(originalfft) / (CHUNK / 2)
fftSpec = fftSpec[:int(CHUNK / 2)]
xf = 1.0 * np.arange(0, FS / 2., FS / (1.*CHUNK))
# Spectrogram
img_array = np.roll(img_array, -1, 0)
img_array[-1:] = 10.0 * np.log10(fftSpec)
# Plotting Graphs
origWave.plot(orig, clear=True)
fftItem.plot(xf, fftSpec, clear=True)
specItem.setImage(img_array, autoLevels=False)
QtGui.QApplication.processEvents()
stream.stop_stream()
stream.close()
audio.terminate()
file.close()
print('Writing wav data')
waveFile = wave.open('recorded.wav', 'wb')
waveFile.setnchannels(CHANNELS)
waveFile.setsampwidth(audio.get_sample_size(FORMAT))
waveFile.setframerate(FS)
waveFile.writeframes(b''.join(frames))
waveFile.close()
print('Finished writing wav data')
print('Exiting...')
sys.exit()
import scipy as sp
from scipy.io.wavfile import read, write
import pylibpd as pd
num_chans = 1
sampling_rate = 44100
# open a Pure Data patch
m = pd.PdManager(num_chans, num_chans, sampling_rate, 1)
patch = pd.libpd_open_patch("ring_mod.pd")
# get the default frame size
frame_size = pd.libpd_blocksize()
# read audio file
audio = read("drums.wav")[1]
# process each frame
out = sp.array([], dtype=sp.int16)
for i in range(0, len(audio), frame_size):
f = audio[i:i + frame_size]
p = m.process(f)
p = sp.fromstring(p, sp.int16)
out = sp.hstack((out, p))
# close the patch
pd.libpd_close_patch(patch)
# write the audio file to disk
write("drums_ringmod.wav", 44100, out)
#!/usr/bin/env python # -*- coding: utf-8 -*- import scipy as sp import scipy.signal as sig import wave fsamp = 44100.0 fpass = 5000.0 fstop = 6000.0 wp = fpass / (fsamp / 2) ws = fstop / (fsamp / 2) b, a = sig.iirdesign(wp, ws, 1, 30) fs, h = sig.freqs(b, a) filename = 'white_noise2.wav' wf = wave.open(filename, 'rb') n = wf.getnframes() s = wf.readframes(n) x = sp.fromstring(s, sp.int16) y = sig.lfilter(b, a, x) o_filename = 'filtered.wav' wf_o = wave.open(o_filename, 'wb') wf_o.setnchannels(1) wf_o.setsampwidth(2) wf_o.setframerate(44100) wf_o.writeframes(sp.int16(y).tostring()) wf_o.close() wf.close()
#!/usr/bin/env python # -*- coding: utf-8 -*- import scipy as sp import scipy.signal as sig import wave fsamp = 44100.0 fpass = 5000.0 fstop = 6000.0 wp = fpass / (fsamp / 2 ) ws = fstop / (fsamp / 2 ) b,a = sig.iirdesign(wp, ws, 1, 30) fs, h = sig.freqs(b,a) filename='white_noise2.wav' wf = wave.open(filename,'rb') n=wf.getnframes() s=wf.readframes(n) x = sp.fromstring(s,sp.int16) y = sig.lfilter(b,a,x) o_filename='filtered.wav' wf_o=wave.open(o_filename,'wb') wf_o.setnchannels(1) wf_o.setsampwidth(2) wf_o.setframerate(44100) wf_o.writeframes(sp.int16(y).tostring()) wf_o.close() wf.close()
import wave
import struct
from scipy import fromstring, int16
# output.wav /Desktop
wavf = '/Users/itounagamitsu/Desktop/AM_python/output_sin.wav'
wr = wave.open(wavf, 'rb')
# waveファイルが持つ性質を取得
ch = wr.getnchannels()
width = wr.getsampwidth()
fr = wr.getframerate()
fn = wr.getnframes()
print("Channel: ", ch)
print("Sample width: ", width)
print("Frame Rate: ", fr)
print("Frame num: ", fn)
print("Params: ", wr.getparams())
print("Total time: ", 1.0 * fn / fr)
# waveの実データを取得し、数値化
data = wr.readframes(wr.getnframes())
wr.close()
X = fromstring(data, dtype=int16)
import math
import numpy as np
import random
import pickle
import wave
from scipy import fromstring, int16
from scipy.fftpack import fft
batch_size = 100
num_of_kinds = 200
dataset = []
dummy = np.zeros((2, 2752512))
wr = wave.open('./data/test01_ch1.WAV', 'rb')
dummy[0, :] = fromstring(wr.readframes(wr.getnframes()), dtype=int16)
wr = wave.open('./data/test01_ch2.WAV', 'rb')
dummy[1, :] = fromstring(wr.readframes(wr.getnframes()), dtype=int16)
dataset.append(dummy)
dummy = np.zeros((2, 2752512))
wr = wave.open('./data/test02_ch1.WAV', 'rb')
dummy[0, :] = fromstring(wr.readframes(wr.getnframes()), dtype=int16)
wr = wave.open('./data/test02_ch2.WAV', 'rb')
dummy[1, :] = fromstring(wr.readframes(wr.getnframes()), dtype=int16)
dataset.append(dummy)
dummy = np.zeros((2, 2752512))
wr = wave.open('./data/test03_ch1.WAV', 'rb')
dummy[0, :] = fromstring(wr.readframes(wr.getnframes()), dtype=int16)
wr = wave.open('./data/test03_ch2.WAV', 'rb')
list3 = [0, 0, 0, 1, 0, 0, 0, 0]
list4 = [0, 0, 0, 0, 1, 0, 0, 0]
list5 = [0, 0, 0, 0, 0, 1, 0, 0]
list6 = [0, 0, 0, 0, 0, 0, 1, 0]
list7 = [0, 0, 0, 0, 0, 0, 0, 1]
i = 0
j = 1
k = 0
###学習用データセット
while (i <= 315):
while (j <= 44):
wr = wave.open('./sound_' + str(i) + '/output/' + str(j) + '.wav',
'rb')
data = wr.readframes(wr.getnframes())
num_data = fromstring(data, dtype='int16') / 32768.0
sound.append(num_data)
j += 1
i += 45
j = 1
###学習用解答データセット
while (k <= 352):
if (k <= 43):
answer.append(list0)
elif (k >= 44 and k <= 87):
answer.append(list1)
elif (k >= 88 and k <= 131):
answer.append(list2)
elif (k >= 132 and k <= 175):
answer.append(list3)
def clean(self, returnString):
"""converts a string delimited with commas into a numpy array """
return scipy.fromstring(returnString, sep=',')
inPCM.setchannels(ch)
inPCM.setrate(fs)
inPCM.setformat(alsaaudio.PCM_FORMAT_S16_LE)
inPCM.setperiodsize(chunk)
signal = zeros(fftLen, dtype=float)
tmpSound = 0
up = 0
down = 0
max = 0
count = 0
while 1: # Check Baby Crying or not
length, data = inPCM.read()
num_data = fromstring(data, dtype="int16")
signal = roll(signal, -chunk)
signal[-chunk:] = num_data
fftspec = fft(signal)
babySound = abs(fftspec[146] *
signal_scale) # This is may be baby's crying Hz (1000Hz)
# First, check baby's minimum sound
if babySound > 50 and babySound < 150:
totalStartTimer = time.time() # Timer start
totalEndTimer = 0
while totalEndTimer - totalStartTimer < 50: # Check during 50 seconds
totalEndTimer = time.time()
if babySound >= tmpSound: # If recent sound is higher than previous sound, doubt baby's sound
up = 1
def readWave(self, stream):
while True:
#print(stream.read(1024))
num_data = fromstring(stream.read(1024), dtype='int16') / 32768.0
print(num_data)
import wave
from scipy import fromstring, int16
# 90秒分に相当するフレーム数を算出
ch = 2
fr = 44100
width = 2
file = open('original.txt', 'r') #読み込みモードでオープン
string = file.read() #readですべて読み込む
# print(string)
X = fromstring(string, dtype=int16)
print(X)
# 出力データを生成
outf = './test_original.wav'
w = wave.Wave_write(outf)
w.setnchannels(2)
w.setsampwidth(4)
w.setframerate(44100)
w.setnframes(fr * ch)
# w.writeframes(x)
w.close()
# regard a word less confident if it has more than one syllable
if wordConfi/wordCount>confidence_threshold:
confidences[wordPy]=wordConfi/wordCount
timestamps[wordPy]=[wordT0,wordT1]
# extract basic information of the wave file
audio=wave.open(wav_file,'r')
ch = audio.getnchannels()
width = audio.getsampwidth()
fr = audio.getframerate()
fn = audio.getnframes()
data = audio.readframes(fn)
audioContent = fromstring(data, dtype=int16)
# split the wave file
if not os.path.exists('./output'):
os.makedirs('./output')
for name in timestamps:
segment=audioContent[int(timestamps[name][0]*fr*ch):int(timestamps[name][1]*fr*ch)]
outd = struct.pack("h" * len(segment), *segment)
ww = wave.open('./output/'+name+'.wav', 'w')
ww.setnchannels(ch)
ww.setsampwidth(width)
ww.setframerate(fr)
ww.writeframes(outd)
ww.close()
def spectrumAnalyzer():
global fftLen, capture_setting, signal_scale
##########################
# Capture Sound from Mic #
##########################
ch = capture_setting["ch"]
fs = capture_setting["fs"]
chunk = capture_setting["chunk"]
inPCM = alsaaudio.PCM(alsaaudio.PCM_CAPTURE)
inPCM.setchannels(ch)
inPCM.setrate(fs)
inPCM.setformat(alsaaudio.PCM_FORMAT_S16_LE)
inPCM.setperiodsize(chunk)
signal = zeros(fftLen, dtype = float)
##########
# Layout #
##########
app = QtGui.QApplication([])
app.quitOnLastWindowClosed()
mainWindow = QtGui.QMainWindow()
mainWindow.setWindowTitle("Spectrum Analyzer")
mainWindow.resize(800, 300)
centralWid = QtGui.QWidget()
mainWindow.setCentralWidget(centralWid)
lay = QtGui.QVBoxLayout()
centralWid.setLayout(lay)
specWid = pg.PlotWidget(name="spectrum")
specItem = specWid.getPlotItem()
specItem.setMouseEnabled(y = False)
specItem.setYRange(0, 1000)
specItem.setXRange(0, fftLen / 2, padding = 0)
specAxis = specItem.getAxis("bottom")
specAxis.setLabel("Frequency [Hz]")
specAxis.setScale(fs / 2. / (fftLen / 2 + 1))
hz_interval = 500
newXAxis = (arange(int(fs / 2 / hz_interval)) + 1) * hz_interval
oriXAxis = newXAxis / (fs / 2. / (fftLen / 2 + 1))
specAxis.setTicks([zip(oriXAxis, newXAxis)])
lay.addWidget(specWid)
mainWindow.show()
# update
for time in range(100):
length, data = inPCM.read()
num_data = fromstring(data, dtype = "int16")
signal = roll(signal, - chunk)
signal[- chunk :] = num_data
fftspec = fft(signal)
print signal[1800:1900]
specItem.plot(abs(fftspec[1 : fftLen / 2 + 1] * signal_scale), clear = True)
QtGui.QApplication.processEvents()
#!/usr/bin/env python
# -*- coding: utf-8 -*-
import numpy as np
import pyaudio
from scipy import fromstring, int16
from matplotlib import pyplot as pl
CHUNK = 1024
p = pyaudio.PyAudio()
input_device_index = 0
stream = p.open(format=pyaudio.paInt16,
channels=2,
rate=44100,
frames_per_buffer=CHUNK,
input=True)
input = stream.read(CHUNK)
print 'input:' + str(len(input))
num_data = fromstring(input, dtype='int16') / 32768.0
print 'num_data:' + str(len(num_data))
wf2 = wave.open(filename2, "rb")
stream = p.open(format=p.get_format_from_width(wf.getsampwidth()),
channels=wf.getnchannels(),
rate=wf.getframerate(),
output=True)
stream2 = p.open(format=p.get_format_from_width(wf2.getsampwidth()),
channels=wf2.getnchannels(),
rate=wf2.getframerate(),
output=True)
data2 = wf2.readframes(CHUNK)
data = wf.readframes(CHUNK)
num_data = fromstring(data, dtype=int16)
num_data2 = fromstring(data, dtype=int16)
#num_data=left(num_data)
#ただのwav再生
while data != '':
list = np.array(num_data)
list2 = np.array(num_data2)
stream.write(list2)
data = wf.readframes(CHUNK)
data2 = wf.readframes(CHUNK)
num_data = fromstring(data, dtype=int16)
num_data2 = fromstring(data2, dtype=int16)
"""リアルタイム録音再生
while stream.is_active():
try:
def readTxt(self, filePath, fileName):
self.fileDateStr = 170816 # UTC date file was created.
if fileName == 'hylebos1baseline.txt':
fileNum = 1
elif fileName == 'hylebos1_a.txt':
fileNum = 2
elif fileName == 'hylebos1_b.txt':
fileNum = 3
elif fileName == 'hylebos2baseline.txt':
fileNum = 4
elif fileName == 'hylebos2_a.txt':
fileNum = 5
self.fileNum = fileNum # File number in set.
self.descript = fileName # Description of the test.
self.minor = '' # Minor note.
self.major = '' # Major note.
self.scanChCount = 8 # number of channels in each A/D scan.
self.chCount = 8 # number of channels written to the file.
self.n = 4096 # Number of samples in the FFT time series.
self.fs = 4096 # (Hz) FFT sampling frequency.
self.xmitFund = 4 # (Hz) Transmit Square
# wave fundamental frequency.
# Read IP measurements from a text file.
with open(filePath, 'r') as fh:
# Number of lines in the file.
lineCount = self.countLines(fh)
# Rewind the pointer in the file back to the beginning.
fh.seek(0)
# Initialize the packet counter.
p = -1
# Initialize the sample index.
s = -1
# Each file contains a file header of length 10 lines,
# followed by packets. Packets contain (11 + n) lines each.
self.pktCount = int((lineCount) / (1 + self.n))
# Dimension arrays indexed by packet.
self.dimArrays()
self.rCurrentMeas = 0.5 # (Ohm) resistance.
self.rExtraSeries = 0.0 # (Ohm).
# Voltage measurement names.
# 0-indexed by channel number.
self.measStr = [
'currentMeas', 'R1-R2', 'N/A', 'N/A', 'N/A', 'N/A', 'N/A',
'N/A'
]
# Construct arrays using the scipy package.
# 5B amplifier maximum of the input range (V).
# 0-indexed by channel number.
self.In5BHi = sp.array([1, 10, 10, 10, 10, 10, 10, 10])
# 5B amplifier maximum of the output range (V).
# 0-indexed by channel number.
self.Out5BHi = sp.array([5, 5, 5, 5, 5, 5, 5, 5])
# MccDaq board AIn() maximum of the input range (V).
# 0-indexed by channel number.
self.ALoadQHi = sp.array([5, 5, 5, 5, 5, 5, 5, 5])
for lidx, line in enumerate(fh, 1):
# Strip off trailing newline characters.
line = line.rstrip('\n')
if line[0] == '$':
# Increment the packet index.
p += 1
# Packet number
spl = line[2:].split(',')
self.pkt[p] = int(spl[0])
# CPU UTC Date and Time Strings.
self.cpuDTStr[p].d = '170816'
self.cpuDTStr[p].t = '160000.00'
# Translate to datetime object.
self.cpuDT[p] = self.str2DateTime(self.cpuDTStr[p])
# GPS UTC Date and Time Strings,
# and latitude and longitude fixes.
self.gpsDTStr[p].d = '170816'
self.gpsDTStr[p].t = '160000.000'
self.lat[p] = 0.
self.longi[p] = 0.
# Translate to datetime object.
self.gpsDT[p] = self.str2DateTime(self.gpsDTStr[p])
assignArr = sp.array([1, 1, 1, 1, 1, 1, 1, 1])
# Count of measurements clipped on the high end of
# the MccDaq board's input range.
self.clipHi[:, p] = assignArr
# Count of measurements clipped on the low end of
# the MccDaq board's input range.
self.clipLo[:, p] = assignArr
# Mean measurement value over the packet as a
# percentage of the AIn() half range.
self.meanPct[:, p] = assignArr
# (pct) Mean value of sample measurements above
# or equal to the mean.
self.meanUpPct[:, p] = assignArr
# (pct) Mean value of sample measurements below
# the mean.
self.meanDnPct[:, p] = assignArr
# Count of measurements above or equal to the mean.
self.countUp[:, p] = assignArr
# Count of measurements below the mean.
self.countDn[:, p] = assignArr
# Set the sample index to 0 to start.
s = 0
elif line[0] != '*':
# Read in raw voltage values.
self.raw[:, p, s] = (sp.fromstring(line,
dtype=float,
sep=','))
if s == self.n - 1:
# Reset the counter to below zero.
s = -1
else:
# Increment the sample counter for the next read.
s += 1
# After the file has been read, perform some calculations.
self.postRead()
import matplotlib.pyplot as plt
import wave
import glob
import scipy as sp
for fname in glob.glob("CAL500_wav/*.wav"):
print(fname)
wo = wave.open(fname, 'rb')
chunk = 65536
data = sp.fromstring(wo.readframes(chunk), sp.int16)
srate = wo.getframerate()
nFFT = 1024
window = sp.hamming(nFFT)
fig, ax = plt.subplots(1)
fig.subplots_adjust(left=0, right=1, bottom=0, top=1)
ax.set_axis_off()
Pxx, freq, bins, im = ax.specgram(data,
NFFT=nFFT,
Fs=srate,
noverlap=512,
window=window)
plt.savefig(fname.replace('wav', 'png'))
plt.close(fig)
def load(cls,
path,
node_file="throats_cellsThroatsGraph_Nodes.txt",
graph_file="throats_cellsThroatsGraph.txt",
network=None,
voxel_size=None,
return_geometry=False):
r"""
Loads network data from an iMorph processed image stack
Parameters
----------
path : string
The path of the folder where the subfiles are held
node_file : string
The file that describes the pores and throats, the
default iMorph name is: throats_cellsThroatsGraph_Nodes.txt
graph_file : string
The file that describes the connectivity of the network, the
default iMorph name is: throats_cellsThroatsGraph.txt
network : OpenPNM Network Object
The OpenPNM Network onto which the data should be loaded. If no
network is supplied then an empty import network is created and
returned.
voxel_size : float
Allows the user to define a voxel size different than what is
contained in the node_file. The value must be in meters.
return_geometry : Boolean
If True, then all geometrical related properties are removed from
the Network object and added to a GenericGeometry object. In this
case the method returns a tuple containing (network, geometry). If
False (default) then the returned Network will contain all
properties that were in the original file. In this case, the user
can call the ```split_geometry``` method explicitly to perform the
separation.
Returns
-------
If no Network object is supplied then one will be created and returned.
If return_geometry is True, then a tuple is returned containing both
the network and a geometry object.
"""
#
path = Path(path)
node_file = os.path.join(path.resolve(), node_file)
graph_file = os.path.join(path.resolve(), graph_file)
# parsing the nodes file
with open(node_file, 'r') as file:
Np = sp.fromstring(file.readline().rsplit('=')[1],
sep='\t',
dtype=int)[0]
vox_size = sp.fromstring(
file.readline().rsplit(')')[1],
sep='\t',
)[0]
# network always recreated to prevent errors
network = GenericNetwork(Np=Np, Nt=0)
# Define expected properies
network['pore.volume'] = sp.nan
scrap_lines = [file.readline() for line in range(4)]
while True:
vals = file.readline().split('\t')
if len(vals) == 1:
break
network['pore.volume'][int(vals[0])] = float(vals[3])
if 'pore.' + vals[2] not in network.labels():
network['pore.' + vals[2]] = False
network['pore.' + vals[2]][int(vals[0])] = True
if voxel_size is None:
voxel_size = vox_size * 1.0E-6 # file stores value in microns
if voxel_size < 0:
raise (Exception('Error - Voxel size must be specfied in ' +
'the Nodes file or as a keyword argument.'))
# parsing the graph file
with open(graph_file, 'r') as file:
# Define expected properties
network['pore.coords'] = sp.zeros((Np, 3)) * sp.nan
network['pore.types'] = sp.nan
network['pore.color'] = sp.nan
network['pore.radius'] = sp.nan
network['pore.dmax'] = sp.nan
network['pore.node_number'] = sp.nan
# Scan file to get pore coordinate data
scrap_lines = [file.readline() for line in range(3)]
line = file.readline()
xmax = 0.0
ymax = 0.0
zmax = 0.0
node_num = 0
while line != 'connectivity table\n':
vals = sp.fromstring(line, sep='\t')
xmax = vals[1] if vals[1] > xmax else xmax
ymax = vals[2] if vals[2] > ymax else ymax
zmax = vals[3] if vals[3] > zmax else zmax
network['pore.coords'][int(vals[0]), :] = vals[1:4]
network['pore.types'][int(vals[0])] = vals[4]
network['pore.color'][int(vals[0])] = vals[5]
network['pore.radius'][int(vals[0])] = vals[6]
network['pore.dmax'][int(vals[0])] = vals[7]
network['pore.node_number'][int(vals[0])] = node_num
node_num += 1
line = file.readline()
# Scan file to get to connectivity data
scrap_lines.append(file.readline()) # Skip line
# Create sparse lil array incrementally build adjacency matrix
lil = sp.sparse.lil_matrix((Np, Np), dtype=int)
while True:
vals = sp.fromstring(file.readline(), sep='\t', dtype=int)
if len(vals) <= 1:
break
lil.rows[vals[0]] = vals[2:]
lil.data[vals[0]] = sp.ones(vals[1])
# fixing any negative volumes or distances so they are 1 voxel/micron
network['pore.volume'][sp.where(network['pore.volume'] < 0)[0]] = 1.0
network['pore.radius'][sp.where(network['pore.radius'] < 0)[0]] = 1.0
network['pore.dmax'][sp.where(network['pore.dmax'] < 0)[0]] = 1.0
# Add adjacency matrix to OpenPNM network
conns = sp.sparse.triu(lil, k=1, format='coo')
network.update({'throat.all': sp.ones(len(conns.col), dtype=bool)})
network['throat.conns'] = sp.vstack([conns.row, conns.col]).T
network['pore.to_trim'] = False
network['pore.to_trim'][network.pores('*throat')] = True
Ts = network.pores('to_trim')
new_conns = network.find_neighbor_pores(pores=Ts, flatten=False)
extend(network=network, throat_conns=new_conns, labels='new_conns')
for item in network.props('pore'):
item = item.split('.')[1]
arr = sp.ones_like(network['pore.' + item])[0]
arr = sp.tile(A=arr, reps=[network.Nt, 1]) * sp.nan
network['throat.' + item] = sp.squeeze(arr)
network['throat.'+item][network.throats('new_conns')] = \
network['pore.'+item][Ts]
trim(network=network, pores=Ts)
# setting up boundary pores
x_coord, y_coord, z_coord = sp.hsplit(network['pore.coords'], 3)
network['pore.front_boundary'] = sp.ravel(x_coord == 0)
network['pore.back_boundary'] = sp.ravel(x_coord == xmax)
network['pore.left_boundary'] = sp.ravel(y_coord == 0)
network['pore.right_boundary'] = sp.ravel(y_coord == ymax)
network['pore.bottom_boundary'] = sp.ravel(z_coord == 0)
network['pore.top_boundary'] = sp.ravel(z_coord == zmax)
# removing any pores that got classified as a boundary pore but
# weren't labled a border_cell_face
ps = sp.where(~sp.in1d(network.pores('*_boundary'),
network.pores('border_cell_face')))[0]
ps = network.pores('*_boundary')[ps]
for side in ['front', 'back', 'left', 'right', 'top', 'bottom']:
network['pore.' + side + '_boundary'][ps] = False
# setting internal label
network['pore.internal'] = False
network['pore.internal'][network.pores('*_boundary',
mode='not')] = True
# adding props to border cell face throats and from pores
Ts = sp.where(
network['throat.conns'][:,
1] > network.pores('border_cell_face')[0] -
1)[0]
faces = network['throat.conns'][Ts, 1]
for item in network.props('pore'):
item = item.split('.')[1]
network['throat.' + item][Ts] = network['pore.' + item][faces]
network['pore.volume'][faces] = 0.0
# applying unit conversions
# TODO: Determine if radius and dmax are indeed microns and not voxels
network['pore.coords'] = network['pore.coords'] * 1e-6
network['pore.radius'] = network['pore.radius'] * 1e-6
network['pore.dmax'] = network['pore.dmax'] * 1e-6
network['pore.volume'] = network['pore.volume'] * voxel_size**3
network['throat.coords'] = network['throat.coords'] * 1e-6
network['throat.radius'] = network['throat.radius'] * 1e-6
network['throat.dmax'] = network['throat.dmax'] * 1e-6
network['throat.volume'] = network['throat.volume'] * voxel_size**3
return network.project
def load(cls, path,
node_file="throats_cellsThroatsGraph_Nodes.txt",
graph_file="throats_cellsThroatsGraph.txt",
network=None, voxel_size=None, return_geometry=False):
r"""
Loads network data from an iMorph processed image stack
Parameters
----------
path : string
The path of the folder where the subfiles are held
node_file : string
The file that describes the pores and throats, the
default iMorph name is: throats_cellsThroatsGraph_Nodes.txt
graph_file : string
The file that describes the connectivity of the network, the
default iMorph name is: throats_cellsThroatsGraph.txt
network : OpenPNM Network Object
The OpenPNM Network onto which the data should be loaded. If no
network is supplied then an empty import network is created and
returned.
voxel_size : float
Allows the user to define a voxel size different than what is
contained in the node_file. The value must be in meters.
return_geometry : Boolean
If True, then all geometrical related properties are removed from
the Network object and added to a GenericGeometry object. In this
case the method returns a tuple containing (network, geometry). If
False (default) then the returned Network will contain all
properties that were in the original file. In this case, the user
can call the ```split_geometry``` method explicitly to perform the
separation.
Returns
-------
If no Network object is supplied then one will be created and returned.
If return_geometry is True, then a tuple is returned containing both
the network and a geometry object.
"""
#
path = Path(path)
node_file = os.path.join(path.resolve(), node_file)
graph_file = os.path.join(path.resolve(), graph_file)
# parsing the nodes file
with open(node_file, 'r') as file:
Np = sp.fromstring(file.readline().rsplit('=')[1], sep='\t',
dtype=int)[0]
vox_size = sp.fromstring(file.readline().rsplit(')')[1], sep='\t',)[0]
# network always recreated to prevent errors
network = GenericNetwork(Np=Np, Nt=0)
# Define expected properies
network['pore.volume'] = sp.nan
scrap_lines = [file.readline() for line in range(4)]
while True:
vals = file.readline().split('\t')
if len(vals) == 1:
break
network['pore.volume'][int(vals[0])] = float(vals[3])
if 'pore.'+vals[2] not in network.labels():
network['pore.'+vals[2]] = False
network['pore.'+vals[2]][int(vals[0])] = True
if voxel_size is None:
voxel_size = vox_size * 1.0E-6 # file stores value in microns
if voxel_size < 0:
raise(Exception('Error - Voxel size must be specfied in ' +
'the Nodes file or as a keyword argument.'))
# parsing the graph file
with open(graph_file, 'r') as file:
# Define expected properties
network['pore.coords'] = sp.zeros((Np, 3))*sp.nan
network['pore.types'] = sp.nan
network['pore.color'] = sp.nan
network['pore.radius'] = sp.nan
network['pore.dmax'] = sp.nan
network['pore.node_number'] = sp.nan
# Scan file to get pore coordinate data
scrap_lines = [file.readline() for line in range(3)]
line = file.readline()
xmax = 0.0
ymax = 0.0
zmax = 0.0
node_num = 0
while line != 'connectivity table\n':
vals = sp.fromstring(line, sep='\t')
xmax = vals[1] if vals[1] > xmax else xmax
ymax = vals[2] if vals[2] > ymax else ymax
zmax = vals[3] if vals[3] > zmax else zmax
network['pore.coords'][int(vals[0]), :] = vals[1:4]
network['pore.types'][int(vals[0])] = vals[4]
network['pore.color'][int(vals[0])] = vals[5]
network['pore.radius'][int(vals[0])] = vals[6]
network['pore.dmax'][int(vals[0])] = vals[7]
network['pore.node_number'][int(vals[0])] = node_num
node_num += 1
line = file.readline()
# Scan file to get to connectivity data
scrap_lines.append(file.readline()) # Skip line
# Create sparse lil array incrementally build adjacency matrix
lil = sp.sparse.lil_matrix((Np, Np), dtype=int)
while True:
vals = sp.fromstring(file.readline(), sep='\t', dtype=int)
if len(vals) <= 1:
break
lil.rows[vals[0]] = vals[2:]
lil.data[vals[0]] = sp.ones(vals[1])
# fixing any negative volumes or distances so they are 1 voxel/micron
network['pore.volume'][sp.where(network['pore.volume'] < 0)[0]] = 1.0
network['pore.radius'][sp.where(network['pore.radius'] < 0)[0]] = 1.0
network['pore.dmax'][sp.where(network['pore.dmax'] < 0)[0]] = 1.0
# Add adjacency matrix to OpenPNM network
conns = sp.sparse.triu(lil, k=1, format='coo')
network.update({'throat.all': sp.ones(len(conns.col), dtype=bool)})
network['throat.conns'] = sp.vstack([conns.row, conns.col]).T
network['pore.to_trim'] = False
network['pore.to_trim'][network.pores('*throat')] = True
Ts = network.pores('to_trim')
new_conns = network.find_neighbor_pores(pores=Ts, flatten=False)
extend(network=network, throat_conns=new_conns, labels='new_conns')
for item in network.props('pore'):
item = item.split('.')[1]
arr = sp.ones_like(network['pore.'+item])[0]
arr = sp.tile(A=arr, reps=[network.Nt, 1])*sp.nan
network['throat.'+item] = sp.squeeze(arr)
network['throat.'+item][network.throats('new_conns')] = \
network['pore.'+item][Ts]
trim(network=network, pores=Ts)
# setting up boundary pores
x_coord, y_coord, z_coord = sp.hsplit(network['pore.coords'], 3)
network['pore.front_boundary'] = sp.ravel(x_coord == 0)
network['pore.back_boundary'] = sp.ravel(x_coord == xmax)
network['pore.left_boundary'] = sp.ravel(y_coord == 0)
network['pore.right_boundary'] = sp.ravel(y_coord == ymax)
network['pore.bottom_boundary'] = sp.ravel(z_coord == 0)
network['pore.top_boundary'] = sp.ravel(z_coord == zmax)
# removing any pores that got classified as a boundary pore but
# weren't labled a border_cell_face
ps = sp.where(~sp.in1d(network.pores('*_boundary'),
network.pores('border_cell_face')))[0]
ps = network.pores('*_boundary')[ps]
for side in ['front', 'back', 'left', 'right', 'top', 'bottom']:
network['pore.'+side+'_boundary'][ps] = False
# setting internal label
network['pore.internal'] = False
network['pore.internal'][network.pores('*_boundary', mode='not')] = True
# adding props to border cell face throats and from pores
Ts = sp.where(network['throat.conns'][:, 1] >
network.pores('border_cell_face')[0] - 1)[0]
faces = network['throat.conns'][Ts, 1]
for item in network.props('pore'):
item = item.split('.')[1]
network['throat.'+item][Ts] = network['pore.'+item][faces]
network['pore.volume'][faces] = 0.0
# applying unit conversions
# TODO: Determine if radius and dmax are indeed microns and not voxels
network['pore.coords'] = network['pore.coords'] * 1e-6
network['pore.radius'] = network['pore.radius'] * 1e-6
network['pore.dmax'] = network['pore.dmax'] * 1e-6
network['pore.volume'] = network['pore.volume'] * voxel_size**3
network['throat.coords'] = network['throat.coords'] * 1e-6
network['throat.radius'] = network['throat.radius'] * 1e-6
network['throat.dmax'] = network['throat.dmax'] * 1e-6
network['throat.volume'] = network['throat.volume'] * voxel_size**3
return network.project
