def apply_dynamic_range_compression(audio, threshold, ratio, attack, release):
return Audio(
sound=pydub_effects.compress_dynamic_range(
audio.sound, threshold, ratio, attack, release
),
old_audio=audio,
)
def addCompression(sound):
#print('play with compression')
threshold = randF(-40, 0)
ratio = randF(0, 10)
attack = randF(0, 10)
release = randF(0, 500)
return compress_dynamic_range(sound, threshold=threshold, ratio=ratio, attack=attack, release=release)
def pydub_dynamic(file, file_ext, out_path, file_name): """ dynamics by pydub """ from pydub import AudioSegment from pydub import effects x = AudioSegment.from_file(file, format=file_ext) x = effects.normalize(x, headroom=0.1) x = effects.compress_dynamic_range(x, threshold=-3, ratio=4.0, attack=5.0, release=50.0) file_handle = x.export(out_path + file_name, format=file_ext)
def record():
numChannels = config.getint('audio', 'channels')
sampleRate = config.getint('audio', 'samplerate')
frameSize = config.getint('audio', 'framesize')
deviceIndex = config.getint('audio', 'deviceIndex')
print('recording...')
audio = pyaudio.PyAudio()
stream = audio.open(format=pyaudio.paInt16,
channels=numChannels,
rate=sampleRate,
input=True,
frames_per_buffer=frameSize,
input_device_index=deviceIndex)
frames = AudioSegment.empty()
while recording:
frames += AudioSegment(stream.read(frameSize),
sample_width=2,
frame_rate=sampleRate,
channels=1)
stream.stop_stream()
stream.close()
audio.terminate()
print('finished recording')
trim = config.getfloat('audio', 'trimSeconds') * 1000
treshold = config.getfloat('audio', 'compressorTreshold')
ratio = config.getfloat('audio', 'compressorRatio')
attack = config.getfloat('audio', 'compressorAttack')
release = config.getfloat('audio', 'compressorRelease')
frames = frames[trim:] # trim begin
frames = frames[:-trim] # trim end
frames = effects.compress_dynamic_range(frames,
threshold=treshold,
ratio=ratio,
attack=attack,
release=release)
frames = effects.normalize(frames) # normalize
if (frames.duration_seconds > 0):
filename = datetime.datetime.now().isoformat()
dataDir = config.get('folders', 'data')
frames.export(os.path.join(dataDir, filename) + '.wav', format='wav')
frames.set_frame_rate(16000).export(os.path.join(dataDir, filename) +
'.flac',
format='flac')
print("wrote", filename, 'duration:', frames.duration_seconds,
'seconds')
def createClipFromRecording(fileName):
fade = config.getfloat('vereinheiter', 'fadeDur') * 1000
silenceLen = int(config.getfloat('vereinheiter', 'silenceLen') * 1000)
silenceTresh = config.getfloat('vereinheiter', 'silenceTresh')
treshold = config.getfloat('vereinheiter', 'compressorTreshold')
ratio = config.getfloat('vereinheiter', 'compressorRatio')
attack = config.getfloat('vereinheiter', 'compressorAttack')
release = config.getfloat('vereinheiter', 'compressorRelease')
frames = AudioSegment.from_wav(os.path.join(recordingsDir, fileName))
if frames.dBFS < config.getfloat('vereinheiter', 'recordingMinDB'):
print('recording level to low:', fileName)
return [None, None]
frames = frames.set_sample_width(4)
frames = frames.fade_in(fade)
frames = frames.fade_out(fade)
frames = effects.normalize(frames)
frames = frames.remove_dc_offset()
nonsilent = silence.detect_nonsilent(frames, silenceLen, silenceTresh)
chunks = [frames[chunk[0]:chunk[1]] for chunk in nonsilent]
frames = AudioSegment.silent(100)
for chunk in chunks:
if len(chunk) > 100:
chunk = chunk.fade_in(20)
chunk = chunk.fade_out(20)
frames = frames.append(AudioSegment.silent(250))
frames = frames.append(chunk)
frames = effects.compress_dynamic_range(frames,
threshold=treshold,
ratio=ratio,
attack=attack,
release=release)
frames = effects.normalize(frames)
if (len(frames) > config.getfloat('vereinheiter', 'minClipLen') * 1000):
clipName = os.path.join(clipsDir, fileName)
frames.export(clipName, format='wav')
einheit = os.path.basename(fileName).split('_')[0]
database.writeClip(einheit, clipName)
print('new clip:', clipName)
return [einheit, clipName]
else:
print('recording too short:', fileName)
return [None, None]
def tools_process_wavs_call(sender, data):
pname = get_value("tools_project_name")
if not pname:
return
# creat sox transformer
tfm = sox.Transformer()
if get_value("tools_trimadd"):
print("Trimming and padding with silence")
set_value("tools_status", "Trimming and padding with silence")
tfm.silence(1, .15, .1)
tfm.silence(-1, .15, .1)
tfm.pad(.25, .5)
if get_value("tools_resample"):
print("Resampling")
set_value("tools_status", "Resampling")
tfm.rate(22050)
if not os.path.exists(pname + '\\processed'):
os.mkdir(pname + '\\processed')
if not os.path.exists(pname + '\\processed\\wavs'):
os.mkdir(pname + '\\processed\\wavs')
with open(pname + "\\output.csv", 'r') as f:
lines = f.readlines()
for line in lines:
wav_path, text = line.split('|')
processedpath = pname + '\\processed\\' + wav_path
text = text.strip() #remove \r\n
tfm.build_file(pname + '\\' + wav_path, processedpath)
print(f"Processing {wav_path}")
set_value("tools_status", "Processing {}".format(wav_path))
if get_value("tools_compress"):
w = AudioSegment.from_wav(processedpath)
w = effects.compress_dynamic_range(w, threshold=-10)
w.export(processedpath, format='wav')
print("Done processing wavs!")
set_value(
"tools_status",
"Done processing wavs. Output at {}/processed/wavs.".format(pname))
print('\a') #system beep