def home():
if request.method == 'GET':
text = "Heya"
else:
text = request.form["words"]
inSound = "static/voice_sans.wav"
outSound = "static/result.wav"
# create 0.1 sec of silence audio segment
space_segment = AudioSegment.silent(
duration=100) #duration in milliseconds
#read wav file to an audio segment
song = AudioSegment.from_wav(inSound)
try:
speedy_song = speedup(song * 6, 10.0, crossfade=5)
except:
speedy_song = song
outVoice = AudioSegment.silent(duration=10)
for character in text:
if not character.isalnum():
if character == " ":
outVoice = outVoice + song + space_segment
else:
outVoice = outVoice + speedy_song
outVoice = speedup(outVoice, 8.0)
outVoice.export(outSound, format="wav")
wave_obj = sa.WaveObject.from_wave_file(outSound)
wave_obj.play()
return render_template("underteller.html", message=text)
def compress_wav(file, max_silence_time):
track = AudioSegment.from_file(file, format='wav')
silences = detect_silence(track, min_silence_len=max_silence_time, silence_thresh=track.dBFS * 2, seek_step=1)
compressed_track = AudioSegment.empty()
prev_cursor = 0
for silence in silences:
compressed_track = compressed_track + track[prev_cursor:silence[0]]
silence_track = track[silence[0]:silence[1]]
speedup_track = speedup(silence_track, playback_speed=(len(silence_track) / max_silence_time), chunk_size=150, crossfade=100)
fade_in = speedup_track.fade(
from_gain=0,
end=0,
duration=250
)
compressed_track = compressed_track + fade_in
prev_cursor = silence[1]
compressed_track = compressed_track + track[prev_cursor:]
compressed_track.export(file, format="wav")
def speedSound(self, sound, keepTime):
# 上一个的长度
beforeLen = 0
left = 100
right = 130
beforeSound = 0
curSound = 0
while left < right:
middle = int((left + right) / 2)
# 参数： 音频 倍数 块大小 过渡时间
curSound = effects.speedup(sound, middle / 100, 300, 100)
soundLen = len(curSound)
if soundLen == keepTime:
return curSound
elif right - left == 1 and soundLen < keepTime and soundLen < beforeLen:
return beforeSound
elif soundLen < keepTime:
right = middle + 1
beforeSound = curSound
beforeLen = soundLen
elif soundLen > keepTime:
left = middle + 1
beforeSound = curSound
beforeLen = soundLen
return curSound
def audioProcessing(sample_list):
procs = []
for s in sample_list:
sample = AudioSegment.from_file(s,s.split('.')[1])
if settings['NORMALIZE']:
sample = normalize(sample)
if settings['SILENCE']:
sample = strip_silence(sample, 1000, silence_thresh=(sample.dBFS-6))
if settings['PADDING']:
sample += AudioSegment.silent(duration=settings['PADDING'])
if settings['PB_SPEED'] != 1:
sample = speedup(sample, settings['PB_SPEED'])
if not settings['CONCAT']:
# Exporting indivudial samples
sample_name = s.split('/')[-1].split('.')[0]
export(sample, sample_name, settings['OUTPUT_FORMAT'], settings['OUTPUT_BITRATE'])
else:
procs.append(sample)
if settings['CONCAT']:
# Exporting sample chain
concatSamples(procs, raw_input('Enter sample chain name: '))
if settings['REM_INDIVIDUAL_SAMPLES']:
removeIndividualSamples(sample_list)
def trackSpeedUp(mSong, iFactor, sTitle, sArtist, sAlbum):
tempSong = speedup(mSong, float(iFactor))
return {
sTitle:
tempSong.export("tempfiles/" + sTitle + ".mp3",
format="mp3",
tags={
"artist": sArtist,
"album": sAlbum
})
}
def speedChange(sound, time):
speed = sound.duration_seconds / time
if speed > 1:
return speedup(sound, speed)
elif speed < 1:
silentt = AudioSegment.silent(
duration=(time - sound.duration_seconds) * 1000)
sound += silentt
return sound
else:
return sound
def get_speak_file(message_content, lang, speed, tld):
pre_processed = BytesIO()
post_processed = BytesIO()
try:
spoken_google = gTTS(message_content, lang=lang, tld=tld)
spoken_google.write_to_fp(fp=pre_processed)
pre_processed.seek(0)
segment = pydub.AudioSegment.from_file(pre_processed,
bitrate=356000,
format="mp3")
segment = effects.speedup(segment, 1.25, 150, 25) # normally 1.25
frames = int(95000 * 1 / speed)
segment.set_frame_rate(frames).export(post_processed, format="s16le")
return post_processed
except AssertionError:
return post_processed
def adjust_audio_length(self, next_segment_start=None):
'''
Adjusts self.audio length to fit a gap between TTS segments.
params: next_segment_start - float, seconds
if next_segment_start=None, audio length is intact
'''
if not next_segment_start or not self.audio:
return
gap = datetime.timedelta(seconds=(next_segment_start - self.start))
l = datetime.timedelta(seconds=self.__len__())
if l > gap:
rate = l / gap
alt = speedup(self.audio, playback_speed=rate)
elif l < gap:
silence = gap - l
silence_duration = silence.total_seconds() * 1000
silence_audio = AudioSegment.silent(duration=silence_duration)
alt = AudioSegment.empty()
alt += self.audio + silence_audio
else:
return
self.audio = alt
def get_pronounces_mp3(self, pronounce_list, playback_speed=2):
filename = ""
for pronounce in pronounce_list:
if pronounce:
filename += pronounce
if "" == filename:
return None
filename += ".mp3"
filepath = os.path.join(self.words_audio_folder,filename)
exist = os.path.isfile(filepath)
if exist:
return filepath
else:
pronounces_mp3 = None
first = True
for pronounce in pronounce_list:
if pronounce:
mp3 = self.get_mp3_pronounce(pronounce)
if mp3:
if first:
pronounces_mp3 = mp3
first = False
else:
pronounces_mp3 = pronounces_mp3 + mp3
if not pronounces_mp3:
return None
else:
pronounces_mp3 = speedup(pronounces_mp3,playback_speed)
# #判断如果超过60秒截断为60秒，微信允许60秒内语音
# seconds = len(pronounces_mp3) / 1000
# if seconds > 60:
# pronounces_mp3 = pronounces_mp3[:1000*60]
# logging.info("{0} play time over 60 seconds".format(filename))
pronounces_mp3.export(filepath, format="mp3")
return filepath
def get_pronounces_mp3(self, pronounce_list, playback_speed=2):
filename = ""
for pronounce in pronounce_list:
if pronounce:
filename += pronounce
if "" == filename:
return None
filename += ".mp3"
filepath = os.path.join(self.words_audio_folder, filename)
exist = os.path.isfile(filepath)
if exist:
return filepath
else:
pronounces_mp3 = None
first = True
for pronounce in pronounce_list:
if pronounce:
mp3 = self.get_mp3_pronounce(pronounce)
if mp3:
if first:
pronounces_mp3 = mp3
first = False
else:
pronounces_mp3 = pronounces_mp3 + mp3
if not pronounces_mp3:
return None
else:
pronounces_mp3 = speedup(pronounces_mp3, playback_speed)
# #判断如果超过60秒截断为60秒，微信允许60秒内语音
# seconds = len(pronounces_mp3) / 1000
# if seconds > 60:
# pronounces_mp3 = pronounces_mp3[:1000*60]
# logging.info("{0} play time over 60 seconds".format(filename))
pronounces_mp3.export(filepath, format="mp3")
return filepath
def speedUp(sound, speed):
return effects.speedup(sound, speed)
def applyEffects(self, track, title):
if random.randint(0, 101) < self.probArray[0]:
track = track * random.randint(2, 5)
print("added repeat effect")
if random.randint(0, 101) < self.probArray[1]:
print("Doing shuffle")
splitted = track[::1000]
splitList = list(splitted)
random.shuffle(splitList)
result = AudioSegment.empty()
for segment in splitList:
result += segment
track = result
# random.shuffle(splitted)
# track = splitted
if random.randint(0, 101) < self.probArray[2]:
print("adding overlay effect")
overlayTrack, new_title = self.prepareNextTrack()
title = title + " + " + new_title
loop = bool(random.getrandbits(1))
position = random.randint(0, len(track))
gain = random.randint(-8, 9)
track = track.overlay(overlayTrack,
position=position,
loop=loop,
gain_during_overlay=gain)
if random.randint(0, 101) < self.probArray[3]:
print("adding fade effect")
fromGain = round(random.uniform(-8, 0), 2)
toGain = round(random.uniform(0.1, 10), 2)
start = random.randint(0, len(track))
end = random.randint(start, len(track) + 1)
track = track.fade(to_gain=toGain,
from_gain=fromGain,
start=start,
end=end)
if random.randint(0, 101) < self.probArray[4]:
print("adding gain effect")
gain = round(random.uniform(-20, 21), 2)
track = track + gain
if random.randint(0, 101) < self.probArray[5]:
print("adding reverse effect")
track = track.reverse()
if random.randint(0, 101) < self.probArray[6]:
print("adding pan effect")
pan = round(random.uniform(-1.0, 1.1), 2)
track = track.pan(pan)
if random.randint(0, 101) < self.probArray[7]:
print("adding invert phase effect")
track = track.invert_phase()
if random.randint(0, 101) < self.probArray[8]:
#Look into this later
print("should be doing lower quality")
track.set_sample_width(1)
track.set_channels(1)
track.set_frame_rate(11025)
if random.randint(0, 101) < self.probArray[9] and len(track) > 150:
print("adding speed effect")
speed = round(random.uniform(1, 2), 2)
print("speed: " + str(speed))
track = speedup(track, playback_speed=speed)
return track, title
def _speedup(self, audio_segment):
print "Speeding up by %s" % self.speed
return speedup(audio_segment, playback_speed=self.speed)
dSpeedRate = sys.argv[4]
strFilePath = sys.argv[5]
strFileName = sys.argv[6]
nListenSpeech = sys.argv[7]
myobj = gTTS(text=strText, lang=strLanguage, slow=False)
if not os.path.exists(strFilePath):
os.makedirs(strFilePath)
#mp3_fp = BytesIO()
#myobj.write_to_fp(mp3_fp)
# Saving the converted audio in a mp3 file named
myobj.save(strFilePath + "\\" + strFileName + ".mp3")
sound = AudioSegment.from_file(strFilePath + "\\" + strFileName + ".mp3")
if float(dSpeedRate) > 1.0:
sound = effects.speedup(sound, float(dSpeedRate))
sound = sound._spawn(sound.raw_data,
overrides={'frame_rate': int(nSampleRate)})
sound.export(strFilePath + "\\" + strFileName + ".mp3",
format="mp3",
bitrate="48k")
# Playing the converted file
if nListenSpeech == "1":
os.system('"' + strFilePath + "\\" + strFileName + ".mp3" + '"')
editFile = audioInput(args.filename)
minDB_silence = -20
num = 0
dBTable = {"low":-25,"moderate":-30,"high":-40}
output = split_on_silence(editFile, 250, dBTable[args.environment_sound], 100,seek_step=1)
try:
outputFile = output[0]
for a in range(1, len(output)):
outputFile += output[a]
#print(str(num))
num+=1
except:
outputFile = editFile
compression = int(100.0* round(1.0 - (float(len(outputFile))/float(len(editFile))),4))
#print("Trying with silence at " + str(minDB_silence) + "dB and the compression is " + str(compression))
exportFile = outputFile
if(args.playback_speed != 1):
newExport = speedup(exportFile, args.playback_speed, 150, 25)
newExport.export("Cut_"+ args.filename, format='mp3')
else:
exportFile.export("Cut_"+ args.filename, format='mp3')
#print("Done. Cut_" + args.filename + " was cut down by " + str(compression) + "%")
print("Cut_"+ args.filename)
def change_speed(self, path, coefficient):
song = AudioSegment.from_file(path)
effects.speedup(song, coefficient).export(path[:len(path) - 4] +
"_speedup.wav",
format="wav")
return "_speedup.wav is ready"
rel_pos += '..' + os.sep
#print('Moving: %s -> %s' % (lastdir, os.getcwd()))
samples = find_samples(name)
#print('Got:', samples)
os.chdir(restore)
return AudioSegment.from_file(rel_pos + samples[0], fmt)
do = my_sample('taiko4.wav') - 2
ko = my_sample('taiko2.wav') - 2
don = my_sample('taiko3.wav') + 2
kon = my_sample('taiko5.wav') + 2
rim = my_sample('rim.wav') - 12
sa = effects.speedup(my_sample('sa.wav')[307:], playback_speed=4.5) - 12
#sa = effects.speedup(my_sample('sa.wav')[310:800], playback_speed=3.5) - 12
shime_do = my_sample('shime1.wav') - 8
shime_ko = my_sample('shime2.wav') - 8
shime_don = my_sample('shime3.wav')
Don = my_sample('taiko1.wav')
#Don = don.overlay(kon)
click = my_sample('click.wav') - 12
char_map = {
'd': do,
'k': ko,
'D': don,
if settings.Debug: print(f"-----Splitting file {file.name}-----")
while end < length:
start_segment = time.perf_counter()
end = min(start + segment_length, length)
if settings.Debug:
print(f"Splitting segment {i}: {start/1000}s - {end/1000}s")
start_action = time.perf_counter()
segment = sound[start:end]
end_action = time.perf_counter()
if settings.Debug:
print(
f"Split complete in {end_action - start_action:0.4f} seconds")
if (settings.PlaybackSpeed > 1):
if settings.Debug: print("Adjusting playback speed")
start_action = time.perf_counter()
segment = effects.speedup(segment,
playback_speed=settings.PlaybackSpeed)
end_action = time.perf_counter()
if settings.Debug:
print(
f"Playback speed ajdusted in {end_action - start_action:0.4f} seconds"
)
start += segment_length - settings.OverlapLen
outFile = "{}{}{}{}{}".format(settings.OutputDir, file.name[:3], "-p",
i, ".mp3")
if settings.Debug: print("Exporting to " + outFile)
start_action = time.perf_counter()
segment.export(outFile, format="mp3")
end_action = time.perf_counter()
if settings.Debug:
print(
