def pack1(self):
if self.width1==1:
fmt="%iB" % self.frames1*self.channel1
else:
fmt="%ih" % self.frames1*self.channel1
out=struct.pack(fmt,*(self.new))
if self.pl1==0 or self.pl1==3:
out_file=wave.open("ampli.wav",'w')
elif self.pl1==1:
out_file=wave.open("wave_mix1.wav","w")
elif self.pl1==2:
out_file=wave.open("wave_mod1.wav",'w')
out_file.setframerate(self.rate1)
out_file.setnframes(self.frames1)
out_file.setsampwidth(self.width1)
out_file.setnchannels(self.channel1)
out_file.writeframes(out)
out_file.close()
if self.pl1==0:
self.read_new("ampli.wav",0)
elif self.pl1==1:
self.read_new("wave_mix1.wav",4)
self.pl1=0
elif self.pl1==2:
self.read_new("wave_mod1.wav",3)
self.pl1=0
else:
self.pl1=0
def mix_files(a, b, c, chann = 2, phase = -1.):
f1 = wave.open(a,'r')
f2 = wave.open(b,'r')
f3 = wave.open(c,'w')
f3.setnchannels(chann)
f3.setsampwidth(2)
f3.setframerate(44100)
f3.setcomptype('NONE','Not Compressed')
frames = min(f1.getnframes(), f2.getnframes())
print "Mixing files, total length %.2f s..." % (frames / 44100.)
d1 = f1.readframes(frames)
d2 = f2.readframes(frames)
for n in range(frames):
if not n%(5*44100): print n // 44100, 's'
if chann < 2:
d3 = struct.pack('h',
.5 * (struct.unpack('h', d1[2*n:2*n+2])[0] +
struct.unpack('h', d2[2*n:2*n+2])[0]))
else:
d3 = ( struct.pack('h',
phase * .3 * struct.unpack('h', d1[2*n:2*n+2])[0] +
.7 * struct.unpack('h', d2[2*n:2*n+2])[0]) +
struct.pack('h',
.7 * struct.unpack('h', d1[2*n:2*n+2])[0] +
phase * .3 * struct.unpack('h', d2[2*n:2*n+2])[0]) )
f3.writeframesraw(d3)
f3.close()
def open_wave(path=None, rw="r"):
global _wavefd
if _wavefd is None:
try:
if path is None:
path = "~/radar" + time.strftime("%Y%m%d_%H%M%S") + ".wav"
path = os.path.expanduser(path)
if rw == "w":
f = open(path, "w") # make the file
f.close()
_wavefd = wave.open(path, rw)
_wavefd.setnchannels(2)
_wavefd.setsampwidth(2)
_wavefd.setframerate(SAMPLING_RATE)
elif rw == "r":
_wavefd = wave.open(path, rw)
try:
assert _wavefd.getnchannels() == 2
assert _wavefd.getsampwidth() == 2
except AssertionError: # SAMPLING_RATE = 48000
print "wave file format is not valid"
print " number channels:%d, sample width:%d" % (_wavefd.getnchannels(), _wavefd.getsampwidth())
raise
if int(_wavefd.getframerate()) != int(SAMPLING_RATE):
print "warning, the sampling rate of the file diagrees with SAMPLING_RATE"
print " file: %d, SAMPLING_RATE: %d" % (int(_wavefd.getframerate()), int(SAMPLING_RATE))
else:
raise ValueError
except:
print "could not open file"
raise # rethrow the exception
else:
raise AssertionError, "a record file is already opened"
def pack2(self):
if self.width2==1:
fmt="%iB" % self.frames2*self.channel2
else:
fmt="%ih" % self.frames2*self.channel2
out=struct.pack(fmt,*(self.new2))
if self.pl2==0 or self.pl2==3:
out_file=wave.open("ampli2.wav",'w')
elif self.pl2==1:
out_file=wave.open("wave_mix2.wav",'w')
elif self.pl2==2:
out_file=wave.open("wave_mod2.wav",'w')
out_file.setframerate(self.rate2)
out_file.setnframes(self.frames2)
out_file.setsampwidth(self.width2)
out_file.setnchannels(self.channel2)
out_file.writeframes(out)
out_file.close()
if self.pl2==0:
self.read_new("ampli2.wav",1)
elif self.pl2==1:
self.read_new("wave_mix2.wav",4)
self.pl2=0
elif self.pl2==2:
self.read_new("wave_mod2.wav",3)
self.pl2=0
else:
self.pl2=0
def noisered(fname):
wr = wave.open(fname, 'r')
par = list(wr.getparams())
par[3] = 0
ww = wave.open('filtered-talk.wav', 'w')
ww.setparams(tuple(par))
lowpass = 200
highpass = 6000
sz = wr.getframerate()
c = int(wr.getnframes()/sz)
for num in range(c):
da = np.fromstring(wr.readframes(sz), dtype=np.int16)
left, right = da[0::2], da[1::2]
lf, rf = np.fft.rfft(left), np.fft.rfft(right)
lf[:lowpass], rf[:lowpass] = 0, 0
lf[55:66], rf[55:66] = 0, 0
lf[highpass:], rf[highpass:] = 0,0
nl, nr = np.fft.irfft(lf), np.fft.irfft(rf)
ns = np.column_stack((nl,nr)).ravel().astype(np.int16)
ww.writeframes(ns.tostring())
wr.close()
ww.close()
fname='filtered-talk.wav'
sptotex(fname)
def clip_wav(wav_file, clip_file, start, stop):
"""Clip from start to stop in wav_file file, save to clip_file.
Args:
wav_file: full path to input wav
clip_file: full path to output wav clip
start: start time in wav_file in milliseconds
stop: stop time in wav_file in milliseconds
"""
win = wave.open(wav_file, 'r')
framerate = win.getframerate()
length = stop - start
frames = int((length / 1000.0) * framerate)
start_frame = int((start / 1000.0) * framerate)
wout = wave.open(clip_file, 'w')
wout.setparams(win.getparams())
print wav_file, clip_file, start, stop
try:
win.setpos(start_frame)
except wave.Error:
print "Bad position"
print
return False
wout.setparams(win.getparams())
wout.writeframes(win.readframes(frames))
win.close()
wout.close()
return True
def splitSong(inputSongName, inputSongFolder, outputSongFolder):
inputSongFileNameNoExt = os.path.splitext(inputSongName)[0]
waveExtension = ".wav"
inputSongFileName = inputSongFolder+'//'+inputSongName
segmentLengthInSeconds = 10;
try:
waveInput = wave.open(inputSongFileName, 'rb')
totalNumberOfFrames = waveInput.getnframes()
frameRate = waveInput.getframerate()
segmentLengthInFrames = (frameRate * segmentLengthInSeconds)-((frameRate * segmentLengthInSeconds)%1024)
numberOfSegments = int(float(totalNumberOfFrames)/segmentLengthInFrames)
#print segmentLengthInFrames
for i in xrange(numberOfSegments):
outputSegmentFileName = outputSongFolder+'//'+inputSongFileNameNoExt + "_part" + str(i) + waveExtension
waveOutput = wave.open(outputSegmentFileName, 'wb')
waveOutput.setparams(waveInput.getparams())
frames = waveInput.readframes(segmentLengthInFrames) # read 10s of input
waveOutput.writeframes(frames) # write 10 s to output segment
waveOutput.close
waveInput.close()
except EOFError as e:
print e
#splitSong("testSong.wav", "//home//christophe//IdeaProjects//GenreClassificationScripts//truncateSong//TestFolderOutput")
def mix_files(a, b, c, chann = 2, phase = -1.):
f1 = wave.open(a, 'r')
f2 = wave.open(b, 'r')
r1, r2 = f1.getframerate(), f2.getframerate()
if r1 != r2:
print("Error: frame rates must be the same!")
sys.exit(1)
f3 = wave.open(c, 'w')
f3.setnchannels(chann)
f3.setsampwidth(2)
f3.setframerate(r1)
f3.setcomptype('NONE', 'Not Compressed')
frames = min(f1.getnframes(), f2.getnframes())
print("Mixing files, total length %.2f s..." % (frames / float(r1)))
d1 = f1.readframes(frames)
d2 = f2.readframes(frames)
for n in range(frames):
if not n % (5 * r1): print(n // r1, 's')
if chann < 2:
d3 = struct.pack('h', int(
.5 * (struct.unpack('h', d1[2*n:2*n+2])[0] +
struct.unpack('h', d2[2*n:2*n+2])[0])))
else:
d3 = ( struct.pack('h', int(
phase * .3 * struct.unpack('h', d1[2*n:2*n+2])[0] +
.7 * struct.unpack('h', d2[2*n:2*n+2])[0])) +
struct.pack('h', int(
.7 * struct.unpack('h', d1[2*n:2*n+2])[0] +
phase * .3 * struct.unpack('h', d2[2*n:2*n+2])[0])) )
f3.writeframesraw(d3)
f3.close()
def main():
HOST = 'localhost'
PORT = 50007
s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
s.bind((HOST, PORT))
s.listen(1)
conn, addr = s.accept()
print 'Connected by', addr
lengthbuf = recvall(conn, 4)
length, = struct.unpack('!I', lengthbuf)
data = recvall(conn, length)
output_framelist = pickle.loads(data)
#conn.close()
ifile = wave.open("../files/9.wav","r")
ofile = wave.open("../files/noise.wav", "w")
ofile.setparams(ifile.getparams())
sampwidth = ifile.getsampwidth()
fmts = (None, "=B", "=h", None, "=l")
fmt = fmts[sampwidth]
dcs = (None, 128, 0, None, 0)
dc = dcs[sampwidth]
for iframe in output_framelist:
oframe = iframe / 2;
oframe += dc
oframe = struct.pack(fmt, oframe)
ofile.writeframes(oframe)
ifile.close()
ofile.close()
def som(self, tipo):
# define stream chunk
chunk = 1024
# open a wav format music
if tipo == 1:
f = wave.open(BEEP, "rb")
elif tipo == 2:
f = wave.open(FIM, "rb")
else:
return
# instantiate PyAudio
p = pyaudio.PyAudio()
# open stream
stream = p.open(format=p.get_format_from_width(f.getsampwidth()),
channels=f.getnchannels(),
rate=f.getframerate(),
output=True)
# read data
data = f.readframes(chunk)
# play stream
while data != "":
stream.write(data)
data = f.readframes(chunk)
# stop stream
stream.stop_stream()
stream.close()
# close PyAudio
p.terminate()
def split_wav_file(filename, part_length, dest_dir=None, basename=None): """ @brief: splits a wav file into smaller @param filename: the name of the file @param part_length: the length in seconds of each part @param dest_dir: the directory in which all parts should be. default is current directory @param basename: the name of the output files. they will be called <basename>_00000.wav @note: the maxium original file length is 833 hours """ if dest_dir is None: dest_dir = '.' if basename is None: basename = os.path.basename(filename) original_file = wave.open(filename, 'r') file_params = original_file.getparams() number_of_frames_per_part = int(original_file.getframerate() * part_length) total_number_of_frames = file_params[3] file_counter = 0 data = 'init' while len(data) != 0: new_filename = basename + '_' + '0' * (DIGITS_IN_NAME-len(str(file_counter))) + str(file_counter) + '.wav' current_file = wave.open(os.path.join(dest_dir, new_filename), 'w') current_file.setparams(file_params) data = original_file.readframes(number_of_frames_per_part) total_number_of_frames -= number_of_frames_per_part current_file.writeframes(data) current_file.close() file_counter += 1 original_file.close()
def save_file(frames):
p = pyaudio.PyAudio() # Create a PyAudio session
if not os.path.isfile(
WAVE_OUTPUT_FILENAME): # If there is not such a file
wf = wave.open(WAVE_OUTPUT_FILENAME, 'wb') # Create the file
wf.setnchannels(CHANNELS) # Set number of channels
wf.setsampwidth(p.get_sample_size(FORMAT)) # Set sampling format
wf.setframerate(RATE) # Set Bit Rate / Frame Rate
wf.writeframes("") # Write nothing
wf.close() # Close the session
wf = wave.open(WAVE_OUTPUT_FILENAME,
'rb') # Open the file with only read permission
n_frames = wf.getnframes() # Get all frames in it
previous_wav = wf.readframes(
n_frames) # Assign all frames to a variable
wf.close() # Close the session
wf = wave.open(WAVE_OUTPUT_FILENAME,
'wb') # Open the file with write permission
wf.setnchannels(CHANNELS) # Set number of channels
wf.setsampwidth(p.get_sample_size(FORMAT)) # Set sampling format
wf.setframerate(RATE) # Set Bit Rate / Frame Rate
wf.writeframes(
previous_wav +
b''.join(frames)) # Write the all frames including previous ones
wf.close() # Close the session
def 切割音檔(被切音檔路徑, 完成音檔指標, 開頭時間, 結尾時間): # 切割一份字格中的一個漢字產生一個音檔 origAudio = wave.open(被切音檔路徑,'r') frameRate = origAudio.getframerate() nChannels = origAudio.getnchannels() sampWidth = origAudio.getsampwidth() nFrames = origAudio.getnframes() # 0.2和0.1是隨便給的 start = int((float(開頭時間) - 0.05)*frameRate) end = int((float(結尾時間) + 0.01)*frameRate) # 確認切割位置：若是輸入的切割位置超出了音檔的長度，就切齊。 if(start < 0): start = 0 if(end > nFrames): end = nFrames anchor = origAudio.tell() origAudio.setpos(anchor + start) chunkData = origAudio.readframes(end-start) # 輸出：存出音檔 chunkAudio = wave.open(完成音檔指標,'w') chunkAudio.setnchannels(nChannels) chunkAudio.setsampwidth(sampWidth) chunkAudio.setframerate(frameRate) chunkAudio.writeframes(chunkData) chunkAudio.close() origAudio.close()
def trs():
card = 'default'
opts, args = getopt.getopt(sys.argv[1:], 'c:')
for o, a in opts:
if o == '-c':
card = a
n1=e1.get()
n2=e2.get()
n3=e3.get()
n4=e3.get()
f = wave.open(n1, 'rb')
A = wave.open(n2, 'rb')
p = wave.open(n3, 'rb')
x = wave.open(n4, 'rb')
sumar(f, A, p,x)
Archivo5 = Reproducir('sumaa.wav')
Archivo5.ruta = "sumaa.wav"
Argumentos = Archivo5.abrir()
Archivo5.inicio(Argumentos[0],Argumentos[1],Argumentos[2])
Archivo5.rep()
def buildFileWav(infiles, files_dir):
import wave
outfile = os.path.join(files_dir,OUTPUT_FILENAME_WAV)
data= []
wav_frames_total = 0
for infile in infiles:
log(infile)
w = wave.open(infile, 'rb')
wav_frames_total += w.getnframes()
data.append( [w.getparams(), w.readframes(w.getnframes())] )
w.close()
output = wave.open(outfile, 'wb')
log(data[0][0])
output.setparams(data[0][0])
# On older (buggy?) Python versions like XBMC's built-in 2.4 .writeframes() seems not to update the "nframes" field of the WAV header
# and the resulting output file is a truncated mess. Therefore, force the nframes for the header and only write raw data.
#
# To give developers even more fun, trying to manually build the full header on python 2.4 is an epic fail:
# - when you read the docs for wave module (python 2.4 and all next versions) it says .setcomptype() takes 2 parameters;
# - when you call .readcomptype() you get a list of two elements;
# - when you feed this list to .setcomptype(), it raises an error that it takes "exactly 3 parameters"!
#
# On a modern Python version, just skip the .setnframes() and inside the loop, call .writeframes() instead of .writeframesraw()
output.setnframes(wav_frames_total)
for i in range(0, 32):
output.writeframesraw(data[i][1])
output.close()
return outfile
def _extract_tokens(self, tokens, output_dir):
if not self.has_audio():
return
filenames = []
with wave.open(self.wav_path,'r') as w_in:
sr = w_in.getframerate()
bitdepth = w_in.getsampwidth()
for t in tokens:
wt = self[t]
name = '{}_{}.wav'.format(self.name,wt.begin)
wt.wav_path = os.path.join(output_dir,name)
filenames.append(wt.wav_path)
if os.path.exists(wt.wav_path):
continue
begpos = int(wt.begin * sr)
endpos = int(wt.end * sr)
duration = endpos - begpos
w_in.setpos(begpos)
data = w_in.readframes(duration)
with wave.open(wt.wav_path,'w') as w_out:
w_out.setnchannels(1)
w_out.setframerate(sr)
w_out.setsampwidth(bitdepth)
w_out.writeframes(data)
return filenames
def mix_files(a, b, c, chann=2, phase=-1.0):
f1 = wave.open(a, "r")
f2 = wave.open(b, "r")
f3 = wave.open(c, "w")
f3.setnchannels(chann)
f3.setsampwidth(2)
f3.setframerate(44100)
f3.setcomptype("NONE", "Not Compressed")
frames = min(f1.getnframes(), f2.getnframes())
print "Mixing files, total length %.2f s..." % (frames / 44100.0)
d1 = f1.readframes(frames)
d2 = f2.readframes(frames)
for n in range(frames):
if not n % (5 * 44100):
print n // 44100, "s"
if chann < 2:
d3 = struct.pack(
"h", 0.5 * (struct.unpack("h", d1[2 * n : 2 * n + 2])[0] + struct.unpack("h", d2[2 * n : 2 * n + 2])[0])
)
else:
d3 = struct.pack(
"h",
phase * 0.3 * struct.unpack("h", d1[2 * n : 2 * n + 2])[0]
+ 0.7 * struct.unpack("h", d2[2 * n : 2 * n + 2])[0],
) + struct.pack(
"h",
0.7 * struct.unpack("h", d1[2 * n : 2 * n + 2])[0]
+ phase * 0.3 * struct.unpack("h", d2[2 * n : 2 * n + 2])[0],
)
f3.writeframesraw(d3)
f3.close()
def chop_input(wav_queue):
while True:
if not input_queue.empty():
handle = wave.open(input_queue.get(), 'rb')
frame_rate = handle.getframerate()
n_frames = handle.getnframes()
window_size = 2 * frame_rate
num_secs = int(math.ceil(n_frames/frame_rate))
snippet_list = []
print "Slicing Audio file..."
#for i in xrange(num_secs):
for i in xrange(38):
'''TODO: USE STRINGIO'''
#wav_buffer = StringIO.StringIO(buffer)
handle2 = '../audio/sample/'+str(i)+'snippet.wav'
snippet = wave.open(handle2 ,'wb')
snippet.setnchannels(2)
snippet.setsampwidth(handle.getsampwidth())
snippet.setframerate(frame_rate)
snippet.writeframes(handle.readframes(window_size))
handle.setpos(handle.tell() - int(1.8 * frame_rate))
snippet.close()
wav_queue.put(handle2)
handle.close()
else:
print "Chop Worker waiting...\n"
sleep(1)
def cchanges(request,a):
slist = audio_story.objects.filter(title=a)
fname_a = slist[0].file_name
stream = wave.open('/home/mayank/Desktop/bep/bep/bep_users/static/story_audio/temp.wav',"rb")
num_channels = stream.getnchannels()
sample_rate = stream.getframerate()
sample_width = stream.getsampwidth()
num_frames = stream.getnframes()
t = stream.readframes(num_frames)
stream.close()
p = wave.open('/home/mayank/Desktop/bep/bep/bep_users/static/story_audio/'+fname_a,"wb")
p.setnchannels(2)
p.setsampwidth(sample_width)
p.setframerate(44100)
p.writeframes(t)
p.close()
story_list = story.objects.filter(title=a)
fname = story_list[0].file_name
fid = open('/home/mayank/Desktop/bep/bep/bep_users/static/story_text/'+fname, 'rb+')
temp = fid.read()
fid.close()
frnd_list = audio_story.objects.filter(title=a)
c = {'temp_title':a,'temp':temp,'frnd_list':frnd_list,'status':'false'}
c.update(csrf(request))
return render_to_response('add_effects.html',c)
def audioFromDrama(dramaName):
# Go to dir containing audio from subs
for path, subdirs, files in os.walk(subAudioDir):
for filename in files:
if filename == str(dramaName + '.wav'):
dramaAudioFileName = os.path.join(
path, filename)
with contextlib.closing(wave.open(
dramaAudioFileName, 'r')) as dramaAudioFile:
# Get wave information from drama wave
rate = dramaAudioFile.getframerate()
channels = dramaAudioFile.getnchannels()
sampleWidth = dramaAudioFile.getsampwidth()
# Make dirs for saving word audio to
wordDir = os.path.join(subIndvAudioDir, word)
if not os.path.exists(wordDir):
os.makedirs(wordDir)
# Prepare a container for wave data
dataContainer = []
# Loop over each instance where the word appears in the drama
for i in range(0, len(db[word][key])):
# Set up new file for writing
# The [i][0] section writes the index number
# taken from the SRT file.
# This will allow future lookup of the sentence
# by index number
fileName = (word + '_' + dramaName + '_' + str(
db[word][key][i][0]) + '.wav')
filePath = os.path.join(wordDir, fileName)
extractAudio = wave.open(filePath, 'w')
extractAudio.setsampwidth(sampleWidth)
extractAudio.setframerate(rate)
extractAudio.setnchannels(channels)
# Set starting point by converting second data from dict
# to frame position for wave
dramaAudioFile.setpos(round((db[word][key][i][1]) * rate))
# Do the same conversion to set the duration of the clip
# add an extra second on just in case
# Note: this could result in an error if the final word
# occurs at the end of a clip. But, with dramas this
# doesn't happen because of music/credits at the end.
second = 1 * rate
duration = round(db[word][key][i][2] * rate + second)
# Collect data from drama audio file
for indvFrames in range(0, duration):
frame = dramaAudioFile.readframes(1)
dataContainer.append(frame)
# Write data to new word audio file
for data in range(0, len(dataContainer)):
extractAudio.writeframesraw(dataContainer[data])
extractAudio.close()
print('Wrote audio file ', fileName)
dataContainer = []
def slice(self, corpus=(), target=(), label=()):
if not corpus: corpus = self.X['corpus_dir']
if not label: label = self.X['result_mlf']
if not target: target = self.X['acoust_dir']
SYS().cldir(target)
A = HTK().readMLF(label)
waveData = dict({})
params = []
for wavefile in A.wav2word:
W = wave.open(corpus + wavefile + '.wav')
scale = float(W.getnframes()) / A.wav2word[wavefile][-1][1]
params = W.getparams()
for word in A.wav2word[wavefile]:
framechunk = W.readframes(int(scale * (word[2] - word[1])))
if word[0] in waveData:
try:
waveData[word[0]] += framechunk
except:
print word[0], 'out of memory'
pass
else:
waveData[word[0]] = framechunk
for word in waveData:
if "<" in word:
continue
# this should only happen for words with illegal characters
S = wave.open(target + word + '.wav', 'w')
S.setparams(params)
S.writeframes(waveData[word])
def slice_wav(in_file, out_file, start_time, stop_time):
# Open file
spf = wave.open(in_file, 'r')
waveform = spf.readframes(-1)
# Adjust for stereo vs. mono encoding
cpf = len(waveform) / spf.getnframes()
adjusted_sample_rate = spf.getframerate() * cpf
start_index = adjusted_sample_rate * start_time
stop_index = adjusted_sample_rate * stop_time
truncated_wave = waveform[start_index:stop_index]
# Create and to new file, preserving most params
f = open(out_file,'a')
f.close()
spf2 = wave.open(out_file,'w')
spf2.setparams(spf.getparams())
spf2.setnframes(len(truncated_wave) / adjusted_sample_rate)
spf2.writeframes(truncated_wave)
spf.close()
spf2.close()
def _locate_and_modify_sounds(self, n_sounds):
"""Find wum samples and preload them as streams"""
sounds = []
for speed in range(1, n_sounds+1): # speed 0 = no sound
filename = os.path.join(self.config['samples'], 'wum' + str(speed) + '.wav')
print(filename)
original_wave = wave.open(filename, 'rb')
original_stream = original_wave.readframes(self.FRAME_RATE * self.FRAME_DURATION)
volumes = []
for volume in range(1, self.MAX_VOLUME + 1):
print("building volume", volume)
fmt = 'h' * (self.FRAME_RATE * self.FRAME_DURATION)
values = struct.unpack(fmt, original_stream)
values = map(lambda sample: int(float(sample) * (float(volume) / self.MAX_VOLUME)) , values)
data_chunk_modified = struct.pack(fmt, *values)
modified_stream = io.BytesIO()
modified_wave = wave.open(modified_stream, 'w')
modified_wave.setparams((self.CHANNELS, self.BPS, self.FRAME_RATE, self.FRAME_RATE * self.FRAME_DURATION, "NONE", "Uncompressed"))
modified_wave.writeframes(data_chunk_modified)
modified_wave.close()
modified_stream.seek(0)
volumes.append(wave.open(modified_stream, 'rb'))
sounds.append(volumes)
#TODO: fail gracefully if a sample is missing
original_wave.close()
return sounds
def add_wav(self,files,read=False) : "files : filename, filename , ...]" # read all files in memory (not too big) and append to chip assert files,'nothing to write' for f in files : # check all BEFORE actual writing fp_in = wave.open(f) print >>sys.stderr,fp_in.getnchannels(), "channels" assert fp_in.getnchannels()!='1',"mono sound file only !" print >>sys.stderr,fp_in.getsampwidth(), "byte width" assert fp_in.getsampwidth()==1,'only 8 bits input !' print >>sys.stderr,fp_in.getframerate(), "samples per second" assert fp_in.getframerate()==8000,'only 8khz samplerate !' self.read_table() for f in files : print >>sys.stderr,'Adding ',f,'...' # read input entirely into memory fp_in = wave.open(f, "r") frameraw = fp_in.readframes(fp_in.getnframes()) # append / prepend ramping sound to avoid clicks pre = ''.join([chr(i) for i in range(0,ord(frameraw[0]),16)]) post = ''.join([chr(i) for i in range(ord(frameraw[-1]),0,-16)]) self.write(f,pre+frameraw+post,read) self.write_table()
def wavSamplesNextFrame(wavFile=None, chunk=None, overlap=None):
#read a frame from wave file back as float numpy array, each index is a channel
#can give chunk/overlap/wavfile name on first call and all is stored in function
if wavSamplesNextFrame.w is None:
if wavFile is None:
sys.exit( "ERROR: must specify WAV FILE!!" )
return
wavSamplesNextFrame.w = wave.open(wavFile, 'r')
wavSamplesNextFrame.name = wavFile
if wavFile is not None:
if (wavFile != wavSamplesNextFrame.name):
wavSamplesNextFrame.w.close()
wavSamplesNextFrame.w = wave.open(wavFile, 'r')
wavSamplesNextFrame.name = wavFile
if chunk is not None:
wavSamplesNextFrame.chunk = chunk
if overlap is not None:
wavSamplesNextFrame.overlap = overlap
#set pointer to wav based on overlap
currentPos = wavSamplesNextFrame.w.tell()
if (currentPos > wavSamplesNextFrame.overlap):
wavSamplesNextFrame.w.setpos(currentPos - wavSamplesNextFrame.overlap)
#read chunk as string
astr = wavSamplesNextFrame.w.readframes(wavSamplesNextFrame.chunk)
# convert binary chunks to short
a = struct.unpack("%ih" % (wavSamplesNextFrame.chunk* wavSamplesNextFrame.w.getnchannels()), astr)
a = [float(val) / pow(2, 15) for val in a]
#make into numpy array by channel
anew = []
for ind in range(wavSamplesNextFrame.w.getnchannels()):
anew.append(a[ind::wavSamplesNextFrame.w.getnchannels()])
return np.array(anew)
def crop_wav(split_start, split_end, input_file_name, output_file_name):
print("splitting")
mydir = os.getcwd()
input_file_path = mydir + "/" + input_file_name
output_file_path = mydir + "/" + output_file_name
input_file = wave.open(input_file_path, 'r')
width = input_file.getsampwidth()
rate = input_file.getframerate()
fpms = rate / 1000 # frames per ms
#programmatically figure out start and end split
length = (split_end - split_start) * fpms
start_index = split_start * fpms
# os.mkdir(path)
output_file = wave.open(output_file_path, "w")
output_file.setparams((input_file.getnchannels(), width, rate, length, input_file.getcomptype(), input_file.getcompname()))
input_file.rewind()
anchor = input_file.tell()
input_file.setpos(anchor + start_index)
output_file.writeframes(input_file.readframes(length))
input_file.close()
output_file.close()
print("finished split")
def slice_wave(in_file, out_file, offset=0, duration=0, max_framerate=None):
"""Write a section of a wavefile to a new file"""
with closing(wave.open(in_file)) as win:
params = win.getparams()
wav = wave.open(out_file, 'wb')
try:
if not max_framerate:
frames, params = _get_frames(in_file, offset, duration)
else:
audio, framerate = get_audio(in_file, offset=offset, duration=duration, max_framerate=max_framerate)
params = list(params)
if len(audio.shape) == 1:
params[0] = 1
else:
params[0] = audio.shape[1]
params[2] = framerate
audio = audio.flatten() + _zeroline[params[1]]
frames = struct.pack(_formats[params[1]] % (len(audio),), *audio)
wav.setparams(params)
wav.writeframes(frames)
finally:
try:
wav.close()
except:
pass
def decrypt(key, cipherfile, decryptfile):
if cipherfile.endswith('.wav'):
# Wave file open
waveRead = wave.open(cipherfile, 'rb')
waveWrite = wave.open(decryptfile, 'wb')
# Reads the parameters
header = waveRead.getparams()
frames = waveRead.getnframes()
sampleWidth = waveRead.getsampwidth()
assert(waveRead.getnchannels() == 1)
ciphertext = [byte for byte in waveRead.readframes(frames)]
plaintext = bytearray([x for x in crypt(key, ciphertext)])
# Writes the parameters and data to wave
waveWrite.setparams(header)
waveWrite.setnchannels(1)
waveWrite.setsampwidth(sampleWidth)
waveWrite.writeframes(plaintext)
waveRead.close()
waveWrite.close()
else:
# Regular file
with open(cipherfile) as cipherTextFile:
with open(decryptfile, mode='w') as plainTextFile:
ciphertext=cipherTextFile.read()
plaintext = ''.join(crypt(key, ciphertext))
plainTextFile.write(plaintext)
def pack3(self):
if self.width3==1:
fmt="%iB" % self.frames3*self.channel3
else:
fmt="%ih" % self.frames3*self.channel3
out=struct.pack(fmt,*(self.new3))
if self.pl3==0 or self.pl3==3:
out_file=wave.open("ampli3.wav",'w')
elif self.pl3==1:
out_file=wave.open("wave_mix3.wav",'w')
elif self.pl3==2:
out_file=wave.open("wave_mod3.wav",'w')
print self.pl3
out_file.setframerate(self.rate3)
out_file.setnframes(self.frames3)
out_file.setsampwidth(self.width3)
out_file.setnchannels(self.channel3)
out_file.writeframes(out)
out_file.close()
if self.pl3==0 :
self.read_new("ampli3.wav",2)
elif self.pl3==1:
self.read_new("wave_mix3.wav",4)
self.pl3=0
elif self.pl3==2:
self.read_new("wave_mod3.wav",3)
self.pl3=0
else:
self.pl3=0
def recognize(data):
s = StringIO.StringIO(data)
s.seek(0)
r = wave.open(s, "rb")
frames = r.readframes(11026)
result = []
while frames != '':
md5 = hashlib.md5(frames).hexdigest()
o = StringIO.StringIO()
w = wave.open(o, "wb")
w.setnchannels(r.getnchannels())
w.setsampwidth(r.getsampwidth())
w.setframerate(r.getframerate())
w.writeframes(frames)
w.close()
o.seek(0)
md5 = hashlib.md5(o.read()).hexdigest()
if md5 in AUDIO_MAP:
result.append(AUDIO_MAP[md5])
else:
(fd, fname) = tempfile.mkstemp(prefix=md5 + "-",
suffix=".wav")
f = os.fdopen(fd, "wb")
o.seek(0)
f.write(o.read())
f.close()
log.warning("Unknown individual character, please add %s" % (
(fname,)))
return None
frames = r.readframes(11026)
return "".join(result)
@author: Harry Ahlas
"""
import os
import wave
import struct
import random
import numpy as np
from numpy import array
import matplotlib.pyplot as plt
os.chdir("C:\\Development\\github\\GAN-tests")
wrd=wave.open("C:\Development\Python\encoder_decoder\encoder_decoder_sounds\Audio\drum_train-06.wav","r")
# Need to improve speed of import. See http://www.cameronmacleod.com/blog/reading-wave-python
def read_whole(filename):
wav_r = wave.open(filename, 'r')
ret = []
while wav_r.tell() < wav_r.getnframes():
decoded = struct.unpack("<h", wav_r.readframes(1))
ret.append(decoded)
return ret
# Import training samples
drum_train_01 = read_whole("Audio\drum_train-01.wav")
drum_train_02 = read_whole("Audio\drum_train-02.wav")
drum_train_03 = read_whole("Audio\drum_train-03.wav")
drum_train_04 = read_whole("Audio\drum_train-04.wav")
data = f.readframes(periodsize)
def usage():
print('usage: playwav.py [-d <device>] <file>', file=sys.stderr)
sys.exit(2)
#Ctrl+C
def signal_handler(sig, frame):
print('Stop: Ctrl+C!')
sys.exit(0)
#Main
if __name__ == '__main__':
signal.signal(signal.SIGINT, signal_handler)
device = 'default'
opts, args = getopt.getopt(sys.argv[1:], 'd:')
#Cerco un se è specificato un altro device
for o, a in opts:
if o == '-d':
device = a
#Come usare
if not args:
usage()
#Read only mode
with wave.open(args[0], 'rb') as f:
play(device, f)
def main():
import sys
import os
import wave
from optparse import OptionParser
import crcmod
from hexdump import hexdump
import binascii
import zlib
usage = "usage: %prog [options] FILENAME"
parser = OptionParser(usage)
parser.add_option("-d",
"--dump",
help="dump configuration to text",
action="store_true",
dest="dump")
parser.add_option("-s",
"--summary",
help="summarize DFU as human readable text",
action="store_true",
dest="summary")
parser.add_option("-o",
"--output",
dest="outfile",
help="write data to OUTFILE")
parser.add_option("-u",
"--unpack",
help="unpack Samples to UNPACK directory",
dest="unpack")
parser.add_option("-p",
"--pack",
help="pack Samples to PACK directory",
dest="pack")
parser.add_option("-r", "--replace",
help="pack REPLACE directory of samples to DFU " + \
"(overwrites contents, either padded or clipped to original size)",
dest="replace")
parser.add_option("-R",
"--raw",
help="use '.raw' sample files (rather than '.wav')",
action="store_true",
dest="raw")
parser.add_option("-t",
"--test",
help="scripted test, dev use only",
action="store_true",
dest="test")
(options, args) = parser.parse_args()
if len(args) != 1:
parser.error("FILE not specified")
print("Opening:", args[0])
infile = open(args[0], "rb")
if not infile:
sys.exit("Unable to open FILE for reading")
else:
data = infile.read()
infile.close()
if data:
config = DFU.parse(data)
if options.summary:
print("Number of samples:", config['block']['total'])
total = 0
btotal = 0
a_count = 1
b_count = 1
for sample in config['block']['samples']:
print("Sample %d-%d: %s (%s bytes, %f sec)" % \
(a_count, b_count, sample['length'], sample['bytes'], int(sample['length'])/32000))
total += int(sample['length'])
btotal += int(sample['bytes'])
if b_count == config['block']['elements'][a_count - 1]:
a_count += 1
b_count = 1
while (a_count <= len(config['block']['elements']) and \
config['block']['elements'][a_count-1] == 0):
a_count += 1
else:
b_count += 1
print("Total length: %d bytes" % btotal)
print("Total length: %f sec" % (total / 32000))
if options.unpack:
path = os.path.join(os.getcwd(), options.unpack)
if os.path.exists(path):
sys.exit("Directory %s already exists" % path)
os.mkdir(path)
a_count = 1
b_count = 1
for data in config['block']['data']:
unpacked = unpack_samples(data.data)
if options.raw:
name = os.path.join(
path, "sample-{0:0=2d}-{1:0=1d}.raw".format(
a_count, b_count))
outfile = open(name, "wb")
for value in unpacked:
outfile.write(value.to_bytes(2, byteorder='little'))
outfile.close()
else:
name = os.path.join(
path, "sample-{0:0=2d}-{1:0=1d}.wav".format(
a_count, b_count))
outfile = wave.open(name, "wb")
outfile.setsampwidth(2)
outfile.setnchannels(1)
outfile.setframerate(32000)
for value in unpacked:
outfile.writeframesraw(
value.to_bytes(2, byteorder='big'))
outfile.close()
if b_count == config['block']['elements'][a_count - 1]:
a_count += 1
b_count = 1
while (a_count <= len(config['block']['elements']) and \
config['block']['elements'][a_count-1] == 0):
a_count += 1
else:
b_count += 1
if options.pack or options.replace:
if options.replace:
path = os.path.join(os.getcwd(), options.replace)
else:
path = os.path.join(os.getcwd(), options.pack)
if not os.path.exists(path):
sys.exit("Directory %s does not exist" % path)
count = 1
a_count = 1
b_count = 1
if options.pack:
# create more than enough empty slots
Empty = Array(
200, Struct(
"length" / Computed(0),
"bytes" / Computed(0),
))
empty = Empty.parse(b"")
config['block']['samples'] = empty
Empty = Array(200, Struct("data" / Bytes(0), ))
empty = Empty.parse(b"")
config['block']['data'] = empty
for sample in config['block']['samples']:
unpacked = []
infile = None
if options.raw:
name = os.path.join(
path, "sample-{0:0=2d}-{1:0=1d}.raw".format(
a_count, b_count))
if os.path.isfile(name):
infile = open(name, "rb")
if infile:
if options.pack:
file_stats = os.stat(name)
length = file_stats.st_size / 2
else:
length = sample['length']
if length > 0xffff:
length = 0xffff
for temp in range(length):
value = infile.read(2)
unpacked.append(
int.from_bytes(value, byteorder='little'))
infile.close()
else:
name = os.path.join(
path, "sample-{0:0=2d}-{1:0=1d}.wav".format(
a_count, b_count))
if os.path.isfile(name):
infile = wave.open(name, "rb")
if infile:
# checks
if infile.getnchannels() != 1:
sys.exit("Samples should be 1 channel: %s" %
name)
if infile.getsampwidth() != 2:
sys.exit("Samples should be 16 bit: %s" % name)
if infile.getframerate() != 32000:
sys.exit("Samples should be 3200KHz: %s" %
name)
if options.pack:
length = infile.getnframes()
else:
length = sample['length']
if length > 0xffff:
length = 0xffff
for temp in range(length):
value = infile.readframes(1)
unpacked.append(
int.from_bytes(value, byteorder='big'))
infile.close()
if len(unpacked):
config['block']['data'][count - 1].data = bytes(
pack_samples(unpacked))
count += 1
if options.pack:
if not infile:
# file not found, advanced to next element
count -= 1
config['block']['elements'][a_count - 1] = b_count - 1
config['block']['total'] = count - 1
a_count += 1
b_count = 1
if a_count == 13:
# all elements done, truncate data arrays
config['block']['samples'] = config['block'][
'samples'][:count - 1]
config['block']['data'] = config['block'][
'data'][:count - 1]
break
else:
config['block']['samples'][count - 2].length = length
config['block']['samples'][count-2].bytes = \
len(config['block']['data'][count-2].data)
b_count += 1
elif b_count == config['block']['elements'][a_count - 1]:
a_count += 1
b_count = 1
while (a_count <= len(config['block']['elements']) and \
config['block']['elements'][a_count-1] == 0):
a_count += 1
else:
b_count += 1
if options.dump:
print(config)
if options.outfile:
# re-calc inner checksum
data = Block.build(config['block'])
crc32 = crcmod.Crc(0x104c11db7,
rev=False,
initCrc=0,
xorOut=0xFFFFFFFF)
crc32.update(data)
config['checksum'] = crc32.crcValue ^ 0xFFFFFFFF
# re-calc outer checksum
data = DFU.build(config)
crc32 = crcmod.Crc(0x104c11db7,
rev=True,
initCrc=0,
xorOut=0xFFFFFFFF)
crc32.update(data)
outfile = open(options.outfile, "wb")
if outfile:
outfile.write(data)
outfile.write(
(crc32.crcValue ^ 0xFFFFFFFF).to_bytes(4,
byteorder='little'))
outfile.close
def getFrequenciesFromPartialAudioFile(waveFilenameOrHandle='temp',
length=10.0,
startSample=0):
'''
It calculates the fundamental frequency at every instant of time of an audio signal
extracted either from the microphone or from an already recorded song.
It uses a period of time defined by the variable "length" in seconds.
It returns a list with the frequencies, a variable with the file descriptor,
and the end sample position.
>>> #_DOCS_SHOW readFile = 'pachelbel.wav'
>>> import os #_DOCS_HIDE
>>> sp = common.getSourceFilePath() #_DOCS_HIDE
>>> readFile = os.path.join(sp, 'audioSearch', 'test_audio.wav') #_DOCS_HIDE
>>> fTup = audioSearch.getFrequenciesFromPartialAudioFile(readFile, length=1.0)
>>> frequencyList, pachelbelFileHandle, currentSample = fTup
>>> for i in range(5):
... print(frequencyList[i])
143.627689055
99.0835452019
211.004784689
4700.31347962
767.827403482
>>> print(currentSample) # should be near 44100, but probably not exact
44032
Now read the next 1 second...
>>> fTup = audioSearch.getFrequenciesFromPartialAudioFile(pachelbelFileHandle, length=1.0,
... startSample=currentSample)
>>> frequencyList, pachelbelFileHandle, currentSample = fTup
>>> for i in range(5):
... print(frequencyList[i])
187.798213268
238.263483185
409.700397349
149.958733396
101.989786226
>>> print(currentSample) # should be exactly double the previous
88064
'''
if "numpy" in base._missingImport:
raise AudioSearchException(
"Cannot run getFrequenciesFromPartialAudioFile without numpy installed"
)
import numpy
if waveFilenameOrHandle == 'temp':
waveFilenameOrHandle = environLocal.getRootTempDir(
) + os.path.sep + 'temp.wav'
if common.isStr(waveFilenameOrHandle):
# waveFilenameOrHandle is a filename
waveFilename = waveFilenameOrHandle
try:
waveHandle = wave.open(waveFilename, 'r')
except IOError:
raise AudioSearchException(
"Cannot open %s for reading, does not exist" % waveFilename)
else:
# waveFilenameOrHandle is a filehandle
waveHandle = waveFilenameOrHandle
storedWaveSampleList = []
environLocal.printDebug("* reading file from disk a part of the song")
for i in range(
int(math.floor(length * recordSampleRate / audioChunkLength))):
startSample = startSample + audioChunkLength
if startSample < waveHandle.getnframes():
data = waveHandle.readframes(audioChunkLength)
storedWaveSampleList.append(data)
freqFromAQList = []
for data in storedWaveSampleList:
samps = numpy.fromstring(data, dtype=numpy.int16)
freqFromAQList.append(autocorrelationFunction(samps, recordSampleRate))
endSample = startSample
return (freqFromAQList, waveHandle, endSample)
import os
import numpy as np
_format = pyaudio.paInt16
_channels = 2
_rate = 44100
_chunk = 2048
maxValue = 2**16
bars = 50
if len(sys.argv) < 2:
print("Syntax: %s entrada.wav" % sys.argv[0])
sys.exit(-1)
wf = wave.open(sys.argv[1], 'rb')
audio = pyaudio.PyAudio()
np.seterr(all='ignore')
stream = audio.open(format=_format,
channels=_channels,
rate=_rate,
output=True)
os.system("clear")
data = wf.readframes(_chunk)
def render(self, cap, output=None, with_audio=False):
"""
Iterate each video frame to print a set of ascii chars
This method reads each video frame from a opencv video capture
resizing the frame and truncate the width if necessary to
print correctly the final string built with the method
convert_frame_pixels_to_ascii.
Finally each final string is printed correctly, if the process
was done too fast will sleep the necessary time to comply
with the fps expected (30 fps by default).
Args:
cap: An OpenCV video capture
output: If the render should be exported to a bash file
"""
v_width = cap.get(cv2.CAP_PROP_FRAME_WIDTH)
v_height = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)
length = cap.get(cv2.CAP_PROP_FRAME_COUNT)
fps = cap.get(cv2.CAP_PROP_FPS)
fps = fps or 30
if with_audio:
import pyaudio
import wave
temp_dir = tempfile.gettempdir()
temp_file_path = temp_dir + "/temp-audiofile-for-vta.wav"
wave_file = wave.open(temp_file_path, 'rb')
chunk = int(wave_file.getframerate() / fps)
p = pyaudio.PyAudio()
stream = p.open(format=p.get_format_from_width(
wave_file.getsampwidth()),
channels=wave_file.getnchannels(),
rate=wave_file.getframerate(),
output=True)
data = wave_file.readframes(chunk)
if output is not None:
file = open(output, 'w+')
file.write("#!/bin/bash \n")
file.write("echo -en '\033[2J' \n")
file.write("echo -en '\u001b[0;0H' \n")
time_delta = 1. / fps
counter = 0
if PLATFORM:
sys.stdout.write("echo -en '\033[2J' \n")
else:
sys.stdout.write('\033[2J')
# read each frame
while cap.isOpened():
t0 = time.process_time()
if PLATFORM:
rows, cols = os.popen('stty size', 'r').read().split()
else:
cols, rows = os.get_terminal_size()
_ret, frame = cap.read()
if frame is None:
break
if with_audio:
data = wave_file.readframes(chunk)
stream.write(data)
# sleep if the process was too fast
if output is None:
if PLATFORM:
sys.stdout.write('\u001b[0;0H')
else:
sys.stdout.write("\x1b[0;0H")
# scale each frame according to terminal dimensions
resized_frame = self.resize_frame(frame, (cols, rows))
# convert frame pixels to colored string
msg = self.convert_frame_pixels_to_ascii(
resized_frame, (cols, rows))
t1 = time.process_time()
delta = time_delta - (t1 - t0)
if delta > 0:
time.sleep(delta)
sys.stdout.write(msg) # Print the final string
else:
print(self.build_progress(counter, length))
if PLATFORM:
print("\u001b[2A")
else:
print("\x1b[2A")
resized_frame = self.resize_frame(frame)
msg = self.convert_frame_pixels_to_ascii(resized_frame,
new_line_chars=True)
file.write("sleep 0.033 \n")
file.write("echo -en '" + msg + "'" + "\n")
file.write("echo -en '\u001b[0;0H' \n")
counter += 1
if with_audio:
stream.close()
p.terminate()
if PLATFORM:
sys.stdout.write("echo -en '\033[2J' \n")
else:
os.system('cls') or None
import pyaudio
import wave
import sys
filename = sys.argv[1]
# set the chunk size of 1024 samples
chunk = 1024
# open the audio file
wf = wave.open(filename, "rb")
# initialize PyAudio object
p = pyaudio.PyAudio()
# open stream object
stream = p.open(format=p.get_format_from_width(wf.getsampwidth()),
channels=wf.getnchannels(),
rate=wf.getframerate(),
output=True)
# read data in chunks
data = wf.readframes(chunk)
# writing to the stream (playing audio)
while data:
stream.write(data)
data = wf.readframes(chunk)
# close stream
stream.close()
import numpy as np
from scipy.io.wavfile import write
import wave
import binascii
import bitarray
import TA.pesan
ba = bitarray.bitarray()
spf = wave.open('testgrde2.wav', 'r')
# Extract Raw Audio from Wav File
signal = spf.readframes(-1)
signal = np.fromstring(signal, 'uint8')
size = signal.size
mod = divmod(size, 4)
headlen = size - mod[1]
# print(size)
# print(mod)
head = signal[:headlen]
tail = signal[headlen:]
print(signal[10467])
pair = head.reshape((-1, 4))
with open("lmgrde2", mode='rb') as file: # b is important -> binary
fileContent = file.read()
fileContent = list(fileContent)
def write_wav(path, params, raw_audio):
with wave.open(path, 'wb') as out:
out.setparams(params)
out.writeframes(raw_audio)
amplitude = []
for i in range(len(amp0)):
amp_per_sec += ((amp0[i])**2)
amp_per_sec += ((amp1[i])**2)
if (i != 0 and i % 500 == 0):
amp_per_sec = (amp_per_sec / 1000)**(1 / 2)
amplitude.extend([amp_per_sec])
amp_per_sec = 0
if (i == len(amp0)):
amp_per_sec = (amp_per_sec / (i % 1000))**(1 / 2)
amplitude.extend([amp_per_sec])
# length (time) of wav file
fname = song
with contextlib.closing(wave.open(fname, 'r')) as f:
frames = f.getnframes()
rate = f.getframerate()
duration = frames / float(rate)
ANALYSISRATE = 88
#raw data from taps.py
Final_Times_Array = [
0.8732759475708008, 1.9280156294504804, 2.88275531133016, 3.88749499320984
]
ind = 0
count = 0
success = True
class VAD(object):
"""docstring for VAD"""
def __init__(self, frame_size = 256, over_lap = 128):
super(VAD, self).__init__()
self.frame_size = frame_size
self.over_lap = over_lap
# method 1: absSum
def get_volume(self,wave_data):
wave_data = np.fromstring(wave_data, dtype=np.int16)
wave_data = wave_data*1.0/max(abs(wave_data)) # normalization
wlen = len(wave_data)
step = self.frame_size - self.over_lap
frame_num = int(math.ceil(wlen*1.0/step))
volume = np.zeros(frame_num)
for i in range(frame_num):
cur_frame = wave_data[np.arange(i*step,min(i*step+self.frame_size,wlen))]
cur_frame = cur_frame - np.median(cur_frame) # zero-justified
volume[i] = np.sum(np.abs(cur_frame))
return volume
# method 2: 10 times log10 of square sum
def get_volumeDB(self,wave_data):
wave_data = np.fromstring(wave_data, dtype=np.int16)
wave_data = wave_data*1.0/max(abs(wave_data)) # normalization
wlen = len(wave_data)
step = self.frame_size - self.over_lap
frame_num = int(math.ceil(wlen*1.0/step))
volume = np.zeros(frame_num)
for i in range(frame_num):
cur_frame = wave_data[np.arange(i*step,min(i*step+self.frame_size,wlen))]
cur_frame = cur_frame - np.mean(cur_frame) # zero-justified
volume[i] = 10*np.log10(np.sum(cur_frame*cur_frame))
return volume
@staticmethod
def string_split(string,width):
return [string[x:x+width] for x in range(0,len(string),width)]
def remove_mute(self,input_file,output_file,case = 1):
wf = wave.open(input_file,'r')
params = wf.getparams()
nchannels, sampwidth, framerate, nframes = params[:4]
wave_data = wf.readframes(nframes)
wf.close()
volume = self.get_volume(wave_data)
if case == 1:
threshhold = max(volume)*0.10
if case == 2:
threshhold = min(volume)*10.0
if case == 3:
threshhold = max(volume)*0.05+min(volume)*5.0
wave_data = self.string_split(wave_data,self.frame_size)
length = len(wave_data)
# print len(volume)
# print len(wave_data)
frames = []
start = 0
end = length-1
for i in range(len(volume)):
if(volume[i]-threshhold)>0:
start = i
break
while(end > start):
if (volume[end]-threshhold)<0:
break
else:
end = end - 1
print start
print end
for i in xrange(start,end):
frames.append(wave_data[i])
wf = wave.open(output_file,"wb")
wf.setnchannels(nchannels)
wf.setsampwidth(sampwidth)
wf.setframerate(framerate)
wf.writeframes(b''.join(frames))
import wave
import pyaudio
import numpy
import pylab
#打开WAV文档，文件路径根据需要做修改
wf = wave.open("D:\\Python\\wavs\\Do-piano-2.20s.wav", "rb")
#创建PyAudio对象
p = pyaudio.PyAudio()
stream = p.open(format=p.get_format_from_width(wf.getsampwidth()),
channels=wf.getnchannels(),
rate=wf.getframerate(),
output=True)
nframes = wf.getnframes()
framerate = wf.getframerate()
def read_duration(fname):
with contextlib.closing(wave.open(fname, 'r')) as f:
frames = f.getnframes()
rate = f.getframerate()
duration = frames / float(rate)
return duration
print("* recording")
frames = []
for i in range(0, int(RATE / CHUNK * RECORD_SECONDS)):
data = stream.read(CHUNK)
frames.append(data) # 2 bytes(16 bits) per channel
print("* done recording")
stream.stop_stream()
stream.close()
p.terminate()
wf = wave.open(WAVE_OUTPUT_FILENAME, 'wb')
wf.setnchannels(CHANNELS)
wf.setsampwidth(p.get_sample_size(FORMAT))
wf.setframerate(RATE)
wf.writeframes(b''.join(frames))
wf.close()
wf = wave.open(WAVE_OUTPUT_FILENAME, 'rb')
p = pyaudio.PyAudio()
stream = p.open(format=p.get_format_from_width(wf.getsampwidth()),
channels=wf.getnchannels(),
rate=wf.getframerate(),
def record_audio(self, audio_file_name, detect_silence=True):
"""
Records audio from the model to the given audio file.
:param audio_file_name:
the wav file to be created with the recorded audio;
recorded with 8-bit linear compression at 8.0 kHz sampling rate
:return:
True if a message was saved.
"""
if self.config["DEBUG"]:
print("> Recording {}...".format(audio_file_name))
self._serial.cancel_read()
with self._lock:
try:
if not self._send(ENTER_VOICE_MODE):
raise RuntimeError("Failed to put modem into voice mode.")
if not self._send(SET_VOICE_COMPRESSION):
raise RuntimeError("Failed to set compression method and sampling rate specifications.")
if not self._send(DISABLE_SILENCE_DETECTION):
raise RuntimeError("Failed to disable silence detection.")
if not self._send(TELEPHONE_ANSWERING_DEVICE_OFF_HOOK):
raise RuntimeError("Unable put modem (TAD) off hook.")
if not self._send(SEND_VOICE_TONE_BEEP):
raise RuntimeError("Failed to play 1.2 second beep.")
if not self._send(ENTER_VOICE_RECIEVE_DATA_STATE, "CONNECT"):
raise RuntimeError("Error: Unable put modem into voice receive mode.")
except RuntimeError as error:
print("Modem initialization error: ", error)
return False
# Record Audio File
start_time = datetime.now()
CHUNK = 1024
audio_frames = []
# Define the range of amplitude values that are to be considered silence.
# In the 8-bit audio data, silence is \0x7f or \0x80 (127.5 rounded up or down)
threshold = 1
min_silence = 127 - threshold
max_silence = 128 + threshold
silent_frame_count = 0
success = True
while 1:
# Read audio data from the Modem
audio_data = self._serial.read(CHUNK)
# Scan the audio data for DLE codes from modem
if (DCE_END_VOICE_DATA_TX in audio_data):
# <DLE><ETX> is in the stream
print(">> <DLE><ETX> Char Recieved... Stop recording.")
break
if (DCE_PHONE_OFF_HOOK in audio_data):
# <DLE>H is in the stream
print(">> Local phone off hook... Stop recording")
break
if (DCE_BUSY_TONE in audio_data):
print(">> Busy Tone... Stop recording.")
break
# Test for silence
if detect_silence:
if len(audio_data) == sum(1 for x in audio_data if min_silence <= x <= max_silence):
# Increment number of contiguous silent frames
silent_frame_count += 1
else:
silent_frame_count = 0
# At 8KHz sample rate, 5 secs is ~40K bytes
if silent_frame_count > 40: # 40 frames is ~5 secs
# TODO: Consider trimming silent tail from audio data.
print(">> Silent frames detected... Stop recording.")
break
# Timeout
if ((datetime.now() - start_time).seconds) > REC_VM_MAX_DURATION:
print(">> Stop recording: max time limit reached.")
break
# Add Audio Data to Audio Buffer
audio_frames.append(audio_data)
# Save the file if there is audio
if len(audio_frames) > silent_frame_count:
print(">> Saving audio file.")
with wave.open(audio_file_name, 'wb') as wf:
wf.setnchannels(1)
wf.setsampwidth(1)
wf.setframerate(8000)
wf.writeframes(b''.join(audio_frames))
else:
print(">> Skipped saving silent audio.")
success = False
print(">> Recording stopped after {} seconds.".format((datetime.now() - start_time).seconds))
# Clear input buffer before sending commands else its
# contents may interpreted as the cmd's return code
self._serial.reset_input_buffer()
# Send End of Recieve Data state by passing "<DLE>!"
# USR-5637 note: The command returns <DLE><ETX>, but the DLE is stripped
# from the response during the test, so we only test for the ETX.
response = "OK" if self.model == "CONEXANT" else ETX_CODE
if not self._send(DTE_END_VOICE_DATA_RX, response):
print("* Error: Unable to signal end of data receive state")
return success
import wave
import numpy as np
import scipy.signal as sp
from sound_source_separation import wave_writer, wave_loader, specgram
from sound_source_separation.stft import short_term_fourier_transform
with wave.open("sample_noise_mixed.wav") as wav:
f, t, stft_data = short_term_fourier_transform(wav)
amp = np.abs(stft_data)
phase = stft_data / np.maximum(amp, 1.e-20)
n_noise_only = 40000
n_noise_only_frame = np.sum(t < n_noise_only / wav.getframerate())
noise_part_amp = amp[:, :n_noise_only_frame]
noise_amp_mean = np.sqrt(np.mean(np.power(noise_part_amp, 2), axis=1, keepdims=True))
p = 1.0
alpha = 3.3
eps = 0.01 * np.power(amp, p)
processed_amp = np.power(np.maximum(np.power(amp, p) - alpha * np.power(noise_amp_mean, p), eps), 1./p)
processed_amp = processed_amp * phase
_, istft_data = sp.istft(processed_amp, fs=wav.getframerate(), nperseg=512, noverlap=256)
istft_data = istft_data.astype(np.int16)
wave_writer.write("sample_noise_reduced.wav", istft_data, wav.getframerate())
with wave.open("sample_noise_reduced.wav") as wav:
wave_loader.play(wav)
freq = np.fft.fftfreq(RATE * 2, 1.0 /
RATE) #check2 seconds frequency to find the maximum
#freq = freq[:int(len(freq)/2)]
freqPeak = freq[np.where(fft == np.max(fft))[0][0]] + 1
print("peak frequency: %d Hz" % freqPeak)
plt.plot(freq, fft)
# plt.axis([0,4000,None,None])
plt.axis([0, 4000, None, None])
plt.show()
plt.close()
print("Finished Listning...")
# close the stream gracefully
stream.stop_stream()
stream.close()
p.terminate()
#stop Recording
stream.stop_stream()
stream.close()
audio.terminate()
waveFile = wave.open(WAVE_OUTPUT_FILENAME, 'wb')
waveFile.setnchannels(CHANNELS)
waveFile.setsampwidth(audio.get_sample_size(FORMAT))
waveFile.setframerate(RATE)
waveFile.writeframes(b''.join(frames))
waveFile.close()
if (not ubicoustics_model.is_file()):
print("Downloading example_model.hdf5 [867MB]: ")
wget.download(MODEL_URL,MODEL_PATH)
# Load Model
context = ubicoustics.everything
context_mapping = ubicoustics.context_mapping
trained_model = model_filename
other = True
selected_file = '../flusense_raw_audio/_0WKVY0n8aE.wav'
selected_context = 'everything'
print("Using deep learning model: %s" % (trained_model))
model = load_model(trained_model)
graph = tf.get_default_graph()
wf = wave.open(selected_file, 'rb')
context = context_mapping[selected_context]
label = dict()
for k in range(len(context)):
label[k] = context[k]
# Setup Callback
def audio_samples(input, frame_count, time_info, status_flags):
global graph
in_data = wf.readframes(frame_count)
np_wav = np.fromstring(in_data, dtype=np.int16) / 32768.0
x = waveform_to_examples(np_wav, RATE)
predictions = []
with graph.as_default():
if x.shape[0] != 0:
def __duration(self, path):
with contextlib.closing(wave.open(path, 'r')) as f:
frames = f.getnframes()
rate = f.getframerate()
duration = frames / float(rate)
return duration
import sys
import wave
import numpy as np
from gcc_phat import gcc_phat
print("start")
sig1 = wave.open("test2.wav", 'rb')#near
ref1 = wave.open("test3.wav", 'rb')#far
rate = sig1.getframerate()
N = rate
window = np.hanning(N)
while True:
sig = sig1.readframes(N)
if len(sig) != 2 * N:
break
ref = ref1.readframes(N)
sig_buf = np.fromstring(sig, dtype='int16')
ref_buf = np.fromstring(ref, dtype='int16')
print len(sig_buf)
print len(ref_buf)
tau, _ = gcc_phat(sig_buf * window, ref_buf * window, fs=rate, max_tau=1)
print(tau*340000)
# plot_wave_file.py
import struct
import wave
from matplotlib import pyplot
# Specify wave file
wavfile = 'author.wav'
# wavfile = 'decay_cosine_mono.wav'
# wavfile = 'decay_cosine_mono_vibrato.wav'
# wavfile = 'sin01_mono.wav'
# wavfile = 'sin01_mono_vibrato.wav'
print('Play the wave file: {0:s}.'.format(wavfile))
# Open wave file
wf = wave.open( wavfile, 'rb')
# Read wave file properties
RATE = wf.getframerate() # Frame rate (frames/second)
WIDTH = wf.getsampwidth() # Number of bytes per sample
LEN = wf.getnframes() # Signal length
CHANNELS = wf.getnchannels() # Number of channels
print('The file has %d channel(s).' % CHANNELS)
print('The file has %d frames/second.' % RATE)
print('The file has %d frames.' % LEN)
print('The file has %d bytes per sample.' % WIDTH)
# exit()
BLOCKLEN = 1000 # Blocksize
dev_index = 2 # device index found by p.get_device_info_by_index(ii)
wav_output_filename = 'test1.wav' # name of .wav file
audio = pyaudio.PyAudio() # create pyaudio instantiation
# create pyaudio stream
stream = audio.open(format = form_1,rate = samp_rate,channels = chans, \
input_device_index = dev_index,input = True, \
frames_per_buffer=chunk)
print("recording")
frames = []
# loop through stream and append audio chunks to frame array
for ii in range(0, int((samp_rate / chunk) * record_secs)):
data = stream.read(chunk, exception_on_overflow=False)
frames.append(data)
print("finished recording")
# stop the stream, close it, and terminate the pyaudio instantiation
stream.stop_stream()
stream.close()
audio.terminate()
# save the audio frames as .wav file
wavefile = wave.open(wav_output_filename, 'wb')
wavefile.setnchannels(chans)
wavefile.setsampwidth(audio.get_sample_size(form_1))
wavefile.setframerate(samp_rate)
wavefile.writeframes(b''.join(frames))
wavefile.close()
NUM = int(sys.argv[1])
BASEPORT = 9000
PORTS = map(lambda x: str(BASEPORT + x), range(0, NUM))
FILE = "sound.wav"
TOPIC = "AUDIO"
context = zmq.Context()
sockets = []
wf = None
for i, port in enumerate(PORTS):
socket = context.socket(zmq.PUB)
socket.bind("tcp://*:%s" % port)
sockets.append(socket)
wf = wave.open(FILE, 'rb')
CHUNK = 4096
FORMAT = pyaudio.paInt16
CHANNELS = 1
RATE = 44100
p = pyaudio.PyAudio()
framecount = 1
while True:
time.sleep(0.001)
framecount += 1
data = wf.readframes(CHUNK)
for i, socket in enumerate(sockets):
import wave
from sound_source_separation import wave_writer, wave_loader, specgram
from sound_source_separation.noise_generator import white_noise
with wave.open("./CMU_ARCTIC/cmu_us_aew_arctic/wav/arctic_a0003.wav") as wav:
speech_data = wave_loader.load_to_mono_array(wav)
n_speech = wav.getnframes()
n_noise_only = 40000
noise_data = white_noise(n_sample=n_noise_only + n_speech)
mixed_data = noise_data
mixed_data[n_noise_only:] += speech_data
wave_writer.write("sample_noise_mixed.wav",
mixed_data,
f_rate=wav.getframerate())
with wave.open("sample_noise_mixed.wav") as wav:
wave_loader.play(wav)
with wave.open("sample_noise_mixed.wav") as wav:
specgram.show_spectrogram(wav)
audio_data += bytes(audio_array)
def onPlaybackAudioFrameBeforeMixing(self, uid, type1, samples, bytesPerSample, channels, samplesPerSec, buffer1, renderTimeMs, avsync_type):
pass
rtc = agorartc.createRtcEngineBridge()
eventHandler = MyRtcEngineEventHandler(rtc)
rtc.initEventHandler(eventHandler)
# Please input your APP ID here.
rtc.initialize("YOUR_APPID", None, agorartc.AREA_CODE_GLOB & 0xFFFFFFFF)
afo = MyAudioFrameObserver()
rtc.joinChannel("", "myChannelName", "", 0)
rtc.startPreview()
rtc.enableVideo()
agorartc.registerAudioFrameObserver(rtc, afo)
input() # Press any key to come to an end.
agorartc.unregisterAudioFrameObserver(rtc, afo)
# Save the audio data into a wave file.
if channel != -1 and samplewidth != -1 and framerate != -1:
print(">>>>>>>> Start saving audio data.")
with wave.open("sound.wav", "wb") as f:
f.setnchannels(channel)
f.setsampwidth(samplewidth)
f.setframerate(framerate)
f.writeframesraw(audio_data)
rtc.leaveChannel()
rtc.release()
from scipy import signal
import pyaudio
import wave
import Queue
import time
from ctypes import *
C_lib = CDLL("./C_lib.so")#load DLL
q = Queue.Queue()
plt_q = Queue.Queue()
tx_q = q
rx_q = q
#get the raw data
filepath="/home/sjh/Desktop/FM/1.wav"
wf = wave.open(filepath,'rb')
params = wf.getparams()
nchannels,samplewidth,framerate,nframes = params[:4]
data = wf.readframes(nframes)
raw_data = np.fromstring(data,dtype=np.short)
#set the buffer size
frame_duration = 20e-3
sample_rate = params[2]
file_sample_num = params[3]
sdr_rate = 1.92e6
osr = int(sdr_rate/sample_rate)
frame_len = sample_rate * frame_duration
frame_num = np.int16(file_sample_num/frame_len)
frame_len = int(frame_len)
import wave
import struct
import numpy as np
if __name__=='__main__':
data_size=40000
fname="test.wav"
frate=11025.0
wav_file=wave.open(fname,'r')
data=wav_file.readframes(data_size)
wav_file.close()
data=struct.unpack('{n}h'.format(n=data_size), data)
data=np.array(data)
w = np.fft.fft(data)
freqs = np.fft.fftfreq(len(w))
print(freqs.min(),freqs.max())
# (-0.5, 0.499975)
# Find the peak in the coefficients
idx=np.argmax(np.abs(w)**2)
freq=freqs[idx]
freq_in_hertz=abs(freq*frate)
print(freq_in_hertz)
# 439.8975
def sound_meter(self):
freq = 0.0
FORMAT = pyaudio.paInt16
CHANNELS = 2
RATE = 44100
CHUNK = 1024
RECORD_SECONDS = 10
WAVE_OUTPUT_FILENAME = "file.wav"
chunk = 2048
counter = 0
audio = pyaudio.PyAudio()
firebaseObject = firebase.FirebaseApplication(
'https://pukaar-9478a.firebaseio.com/', None)
# start Recording
stream = audio.open(format=FORMAT,
channels=CHANNELS,
rate=RATE,
input=True,
frames_per_buffer=CHUNK)
print("recording...")
frames = []
for i in range(0, int(RATE / CHUNK * RECORD_SECONDS)):
data = stream.read(CHUNK)
frames.append(data)
print("finished recording")
# stop Recording
stream.stop_stream()
stream.close()
audio.terminate()
waveFile = wave.open(WAVE_OUTPUT_FILENAME, 'wb')
waveFile.setnchannels(CHANNELS)
waveFile.setsampwidth(audio.get_sample_size(FORMAT))
waveFile.setframerate(RATE)
waveFile.writeframes(b''.join(frames))
waveFile.close()
# Decibels recognizer
# open up a wave
wf = wave.open('bird.wav', 'rb')
swidth = wf.getsampwidth()
RATE = wf.getframerate()
window = np.blackman(chunk)
# open stream
p = pyaudio.PyAudio()
stream = p.open(format=p.get_format_from_width(wf.getsampwidth()),
channels=wf.getnchannels(),
rate=RATE,
output=True)
# read some data
data = wf.readframes(chunk)
# play stream and find the frequency of each chunk
while len(data) == chunk * swidth:
# unpack the data and times by the hamming window
indata = np.array(
wave.struct.unpack("%dh" %
(len(data) / swidth), data)) * window
# Take the fft and square each value
fftData = abs(np.fft.rfft(indata))**2
# find the maximum
which = fftData[1:].argmax() + 1
# use quadratic interpolation around the max
if which != len(fftData) - 1:
y0, y1, y2 = np.log(fftData[which - 1:which + 2:])
x1 = (y2 - y0) * .5 / (2 * y1 - y2 - y0)
# find the frequency and output it
thefreq = (which + x1) * RATE / chunk
freq = thefreq
if thefreq > 2500:
if counter == 0:
print("First millisecond")
counter = counter + 1
else:
result = firebaseObject.post('/sound', {
'decibels': thefreq,
'tag': "High frequency"
})
print(thefreq)
counter = counter + 1
else:
counter = counter + 1
result = firebaseObject.post('/sound', {
'decibels': thefreq,
'tag': "Low frequency"
})
else:
thefreq = which * RATE / chunk
result = firebaseObject.post('/sound', {'decibels': thefreq})
# read some more data
data = wf.readframes(chunk)
if data:
stream.write(data)
stream.close()
p.terminate()
import numpy
import scipy.io.wavfile as wav
import sys
import wave
#Variable inputs to Silence Detection module
WindowSize = 5 #milliseconds
SilenceThreshold = 350 #milliseconds
try:
#Read wav file
(SamplingRate,Wavdata) = wav.read("../wav/" + sys.argv[1] );
TotalSamples = len(Wavdata);
Wavobj = wave.open("../wav/" + sys.argv[1],'r');
BytesPerSample = Wavobj.getsampwidth();
SamplesPerFrame = int((SamplingRate*WindowSize)/1000); #Samples per frame = (samples/second)*(seconds/frame)
#Normalize data to range [-1,+1)
if BytesPerSample == 1:
Wavdata = (Wavdata-128)/128.0;
elif BytesPerSample == 2:
Wavdata = Wavdata/32768.0;
elif BytesPerSample == 3:
Wavdata = Wavdata/(2.0^23);
elif BytesPerSample == 4:
Wavdata = Wavdata/32768.0; #Why? Shouldn't the scaling factor be 2^32?
if len(sys.argv) < 2:
print("Plays a wave file.\n\nUsage: %s filename.wav volume_level" %
sys.argv[0])
sys.exit(-1)
gain = int(sys.argv[2])
print("gain is = ", str(gain))
multiplier = math.pow(10, (gain / 20))
print("multiplier is = ", multiplier)
#wf = wave.open(sys.argv[1], 'rb')
music = AudioSegment.from_wav(sys.argv[1]) #testestest
music = music + gain
music.export("modified_song_by_setvolume.py.wav", "wav")
wf = wave.open("modified_song_by_setvolume.py.wav", 'rb')
# open stream to play audio through
output_stream = p.open(format=p.get_format_from_width(wf.getsampwidth()),
channels=wf.getnchannels(),
rate=wf.getframerate(),
output=True)
def play():
global write_data, music_avg, gain
#write_data = [min(255, d+gain) for d in wf.readframes(num_frames)]
write_data = [
min(255, int(d * multiplier)) for d in wf.readframes(num_frames)
]
#print("write data: ", str(write_data))
