def __init__(self, config):
self.config = config
self.filenames = librosa.util.find_files(config['data']['sample_set_dir'])
self.set = {}
for file in self.filenames:
filename = os.path.basename(file)
if filename[-1] == '/':
filename = filename[0:-1]
audio, labels = pp.get_samples_and_labels(filename, config)
if config['data']['type'] == 'mel':
spec = torch.Tensor(audio_utils.wav2spectrogram(audio)).t()
melspec = torch.Tensor(audio_utils.wav2melspectrogram(audio)).t()
self.set[filename] = [melspec, labels, spec]
else:
audio = np.array(audio, dtype = np.float64)
f0, ap, sp, coded_sp = pw.cal_mcep(audio)
coded_sp = torch.Tensor(coded_sp.T)
self.set[filename] = [f0, ap, sp, coded_sp, labels]
def _single_conversion(filename, model, one_hot_emo):
'''
THIS WON'T WORK RIGHT NOW, USE THE WORLD CONVERSION LOOP IN MAIN
Call only from __main__ section in this module. Generates sample converted
into each emotion.
(str) filename - name.wav file to be converted
(StarGAN-emo-VC1) model - pretrained model to perform conversion
(torch.Tensor(long)) one_hot_emo - one hot encoding of emotion to convert to
'''
wav, labels = pp.get_wav_and_labels(filenames[5],
config['data']['dataset_dir'])
wav = np.array(wav, dtype=np.double)
f0, ap, sp, coded_sp = preprocess_world.cal_mcep(wav)
coded_sp = coded_sp.T
coded_sp_torch = torch.Tensor(coded_sp).unsqueeze(0).unsqueeze(0).to(
device=device)
fake = model.G(coded_sp_torch, one_hot_emo.unsqueeze(0))
fake = fake.squeeze()
print("Sampled size = ", fake.size())
converted_sp = fake.cpu().detach().numpy()
converted_sp = np.array(converted_sp, dtype=np.float64)
sample_length = converted_sp.shape[0]
if sample_length != ap.shape[0]:
ap = np.ascontiguousarray(ap[0:sample_length, :], dtype=np.float64)
f0 = np.ascontiguousarray(f0[0:sample_length], dtype=np.float64)
f0 = np.ascontiguousarray(f0[20:-20], dtype=np.float64)
ap = np.ascontiguousarray(ap[20:-20, :], dtype=np.float64)
converted_sp = np.ascontiguousarray(converted_sp[40:-40, :],
dtype=np.float64)
coded_sp = np.ascontiguousarray(coded_sp[20:-20, :], dtype=np.float64)
target = np.argmax(one_hot_emo)
out_name = filename[:-4] + str(labels[1]) + "to" + target + ".wav"
audio_utils.save_world_wav([f0, ap, sp, converted_sp],
model.name + '_converted', out_name)
# # coded_sp_temp = np.copy(coded_sp).T
# # print(coded_sp_temp.shape)
# filename_wav = f[0:-4] + "_" + str(int(labels[0].item())) + ".wav"
# print(coded_sp.shape)
# it = str(args.iteration)[0:3]
# audio_utils.save_world_wav([f0,ap,sp,coded_sp], args.out_dir + '_evalSet', filename_wav)
########################################
# WORLD CONVERSION LOOP #
########################################
for file_num, f in enumerate(filenames):
# wav, labels = pp.get_wav_and_labels(f, config['data']['dataset_dir'])
wav = np.array(wav, dtype = np.float64)
labels = np.array(labels)
f0_real, ap_real, sp, coded_sp = preprocess_world.cal_mcep(wav)
# coded_sp_temp = np.copy(coded_sp).T
# print(coded_sp_temp.shape)
coded_sp = coded_sp.T
coded_sp = torch.Tensor(coded_sp).unsqueeze(0).unsqueeze(0).to(device = device)
with torch.no_grad():
# print(emo_targets)
for i in range (0, emo_targets.size(0)):
f0 = np.copy(f0_real)
ap = np.copy(ap_real)
# coded_sp_temp_copy = np.copy(coded_sp_temp)
# coded_sp = np.copy(coded_sp)
f0 = audio_utils.f0_pitch_conversion(f0, (labels[0],labels[1]),