def generatefromwave(wave, name):
maxchannels = wave.getnchannels()
amp = wave.getsampwidth()
raw = wave.readframes(wave.getnframes())
track = []
i = 0
while (i != len(raw)):
value = struct.unpack("<h", raw[i:i + amp])
value = value[0]
#value de -2**15 a 2**15 (ou outros valores dependendo de amp)
value = value / 2**(8 * (amp) - 1)
track.append(value)
i += amp
#TODO:descobrir como o readnbytes me da os canais
#primeira tentativa: assumir que os canais estao concatenados:
channels = []
size = wave.getnframes()
for i in range(maxchannels):
channels.append([])
j = 0
while ((j + 1) * 728) <= len(track):
channels[i].append(track[j * 728:(j + 1) * 728])
j += 1
return AudioTrack(channels, name)
def extractSamples(wave, start, end):
sampleRate = wave.getframerate()
duration = end - start
assert duration > 0
wave.setpos(start * sampleRate)
return [ struct.unpack_from("<h", wave.readframes(1))[0]
for i in range(0, int(duration * sampleRate))]
def get_speech_int_array(wave, start, end):
vad = webrtcvad.Vad(3)
start = max(0, start)
samples_per_second = wave.getframerate()
#print "Framerate: %i" % samples_per_second
samples_per_frame = int(SPEECH_FRAME_SEC * samples_per_second)
total_samples = wave.getnframes()
#print "Samples per frame: %i" % samples_per_frame
wave.rewind()
try:
wave.setpos(start * samples_per_second)
except:
print "faield to set pos %f" % start
wave_view_int = []
while wave.tell() < min(end * samples_per_second, total_samples):
#wave_view_str += "1" if vad.is_speech(wave.readframes(samples_to_get), sample_rate) else "0"
try:
wav_samples = wave.readframes(samples_per_frame)
wave_view_int.append(
1 if vad.is_speech(wav_samples, samples_per_second) else 0)
except Exception as ex:
print("Exception: " + str(ex))
return []
return wave_view_int
def __init__(self, wavfile):
# Load wav file
# Fetch sampling rate adn signal object
wf = w.open(wavfile)
self.sampling_rate = wf.getframerate()
s = w.readframes(w.getnframes())
self.signal = np.frombuffer(s, dtype="int16")
ef.close()
# Define constants
n0 = 0 # start point for sa
N = 1024 # number of sample
def __init__(self, wavfile):
# Load wav file
# Fetch sampling rate adn signal object
wf = w.open(wavfile)
self.sampling_rate = wf.getframerate()
s = w.readframes(w.getnframes())
self.signal = np.frombuffer(s, dtype="int16")
ef.close()
# Define constants
n0 = 0 # start point for sa
N = 1024 # number of sample
def get_speech_int_array(wave, start, end):
vad = webrtcvad.Vad(3)
samples_per_second = wave.getframerate()
samples_per_frame = int(SPEECH_FRAME_SEC * samples_per_second)
wave.setpos(start * samples_per_second)
wave_view_int = []
while wave.tell() < end * samples_per_second:
#wave_view_str += "1" if vad.is_speech(wave.readframes(samples_to_get), sample_rate) else "0"
try:
wav_samples = wave.readframes(samples_per_frame)
wave_view_int.append(
1 if vad.is_speech(wav_samples, samples_per_second) else 0)
except:
return []
return wave_view_int
def get_speech_int_array(wave, start=0, end=-1):
vad = webrtcvad.Vad(0)
samples_per_second = wave.getframerate()
samples_per_frame = int(SPEECH_FRAME_SEC * samples_per_second)
if end == -1:
end = float(wave.getnframes()) / samples_per_second
wave.setpos(start * samples_per_second)
wave_view_int = []
while wave.tell() < end * samples_per_second:
#wave_view_str += "1" if vad.is_speech(wave.readframes(samples_to_get), sample_rate) else "0"
try:
wav_samples = wave.readframes(samples_per_frame)
wave_view_int.append(
1 if vad.is_speech(wav_samples, samples_per_second) else 0)
except:
print "Exception reading frames"
return []
return wave_view_int
def trim_silence(wave, output_file_path):
vad = webrtcvad.Vad(3)
VAD_WINDOW_SEC = 0.01
samples_per_second = wave.getframerate()
samples_per_frame = int(VAD_WINDOW_SEC * samples_per_second)
total_samples = wave.getnframes()
#print('samples_step: %i' % samples_per_frame)
wave_view_str = ""
wave_view_int = []
while wave.tell() < total_samples:
#wave_view_str += "1" if vad.is_speech(wave.readframes(samples_to_get), sample_rate) else "0"
try:
wav_samples = wave.readframes(samples_per_frame)
val = 1 if vad.is_speech(wav_samples, samples_per_second) else 0
wave_view_int.append(val)
wave_view_str += str(val)
#print "current_pos: %i" % wave.tell()
except Exception as ex:
print("Exception: " + str(ex))
return []
print wave_view_str
#Visualizer help from 'courageousillumination'
#https://github.com/courageousillumination/visualizer/blob/master/visualizer.py#L36
import pygame as pgame
import numpy as np
import wave
import array
pgame.mixer.init() #initalizes the mixer element
pgame.init() #initalizes the game screen
pgame.mixer.music.load('ArchieCut.wav')
sound_file = wave.open('ArchieCut.wav')
data = wave.readframes()
print sound_file.getparams()
stringOfBytes = np.fromstring(data, '<H')
#'<H' = little endian unsigned short
#needed to grab the data from the sound
import matplotlib.pyplot as plt
import numpy as np
import wave
import sys
wave = wave.open('imp.wav', 'r')
signal = wave.readframes(-1)
signal = np.fromstring(signal, 'Int16')
plt.plot(signal)
plt.show()
def play(self, wave):
samples = wave.readframes(self.AUDIO_CHUNK)
while samples:
self.stream.write(samples)
samples = wave.readframes(self.AUDIO_CHUNK)
wave.rewind()
