def wavInfoChk(wave, name):
print(name + " # channels : " + str(wave.getnchannels()))
print(name + " sampling : " + str(wave.getframerate()) + " Hz")
print(name + " samp width : " + str(wave.getsampwidth()) + " byte")
print(name + " length : " + str(wave.getnframes()))
if wave.getnchannels() != 1 or wave.getframerate(
) != 16000 or wave.getsampwidth() != 2:
print("ERROR: Input wav files must be Monaural/16kHz/16bit PCM",
file=sys.stderr)
sys.exit(1)
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 play(self):
if self.stream:
self.stream.close()
self.stream = p.open(format=p.get_format_from_width(wf.getsampwidth()),
channels=wave.getnchannels(),
rate=wave.getframerate(),
output=True)
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