diff --git a/go-web.bat b/go-web.bat index 6ade321..db1dec5 100644 --- a/go-web.bat +++ b/go-web.bat @@ -1,2 +1,2 @@ -runtime\python.exe infer-web.py --pycmd runtime\python.exe +runtime\python.exe infer-web.py --pycmd runtime\python.exe --port 7897 pause diff --git a/infer-web.py b/infer-web.py index ae43d05..3347626 100644 --- a/infer-web.py +++ b/infer-web.py @@ -1,5 +1,5 @@ from multiprocessing import cpu_count -import threading +import threading,pdb,librosa from time import sleep from subprocess import Popen from time import sleep @@ -17,6 +17,7 @@ os.environ["TEMP"] = tmp warnings.filterwarnings("ignore") torch.manual_seed(114514) from i18n import I18nAuto +import ffmpeg i18n = I18nAuto() # 判断是否有能用来训练和加速推理的N卡 @@ -235,7 +236,7 @@ def vc_multi( yield traceback.format_exc() -def uvr(model_name, inp_root, save_root_vocal, paths, save_root_ins): +def uvr(model_name, inp_root, save_root_vocal, paths, save_root_ins,agg): infos = [] try: inp_root = inp_root.strip(" ").strip('"').strip("\n").strip('"').strip(" ") @@ -246,6 +247,7 @@ def uvr(model_name, inp_root, save_root_vocal, paths, save_root_ins): save_root_ins.strip(" ").strip('"').strip("\n").strip('"').strip(" ") ) pre_fun = _audio_pre_( + agg=int(agg), model_path=os.path.join(weight_uvr5_root, model_name + ".pth"), device=device, is_half=is_half, @@ -254,10 +256,25 @@ def uvr(model_name, inp_root, save_root_vocal, paths, save_root_ins): paths = [os.path.join(inp_root, name) for name in os.listdir(inp_root)] else: paths = [path.name for path in paths] - for name in paths: - inp_path = os.path.join(inp_root, name) + for path in paths: + inp_path = os.path.join(inp_root, path) + need_reformat=1 + done=0 try: - pre_fun._path_audio_(inp_path, save_root_ins, save_root_vocal) + info = ffmpeg.probe(inp_path, cmd="ffprobe") + if(info["streams"][0]["channels"]==2 and info["streams"][0]["sample_rate"]=="44100"): + need_reformat=0 + pre_fun._path_audio_(inp_path, save_root_ins, save_root_vocal) + done=1 + except: + need_reformat = 1 + traceback.print_exc() + if(need_reformat==1): + tmp_path="%s/%s.reformatted.wav"%(tmp,os.path.basename(inp_path)) + os.system("ffmpeg -i %s -vn -acodec pcm_s16le -ac 2 -ar 44100 %s -y"%(inp_path,tmp_path)) + inp_path=tmp_path + try: + if(done==0):pre_fun._path_audio_(inp_path, save_root_ins, save_root_vocal) infos.append("%s->Success" % (os.path.basename(inp_path))) yield "\n".join(infos) except: @@ -1147,6 +1164,15 @@ with gr.Blocks() as app: ) with gr.Column(): model_choose = gr.Dropdown(label=i18n("模型"), choices=uvr5_names) + agg = gr.Slider( + minimum=0, + maximum=20, + step=1, + label="人声提取激进程度", + value=10, + interactive=True, + visible=False#先不开放调整 + ) opt_vocal_root = gr.Textbox( label=i18n("指定输出人声文件夹"), value="opt" ) @@ -1161,6 +1187,7 @@ with gr.Blocks() as app: opt_vocal_root, wav_inputs, opt_ins_root, + agg ], [vc_output4], ) @@ -1246,7 +1273,7 @@ with gr.Blocks() as app: with gr.Row(): save_epoch10 = gr.Slider( minimum=0, - maximum=200, + maximum=50, step=1, label=i18n("保存频率save_every_epoch"), value=5, diff --git a/infer_uvr5.py b/infer_uvr5.py index b7d484d..07da7eb 100644 --- a/infer_uvr5.py +++ b/infer_uvr5.py @@ -13,7 +13,7 @@ from scipy.io import wavfile class _audio_pre_: - def __init__(self, model_path, device, is_half): + def __init__(self, agg,model_path, device, is_half): self.model_path = model_path self.device = device self.data = { @@ -22,7 +22,7 @@ class _audio_pre_: "tta": False, # Constants "window_size": 512, - "agg": 10, + "agg": agg, "high_end_process": "mirroring", } nn_arch_sizes = [ @@ -139,7 +139,7 @@ class _audio_pre_: wav_instrument = spec_utils.cmb_spectrogram_to_wave(y_spec_m, self.mp) print("%s instruments done" % name) wavfile.write( - os.path.join(ins_root, "instrument_{}.wav".format(name)), + os.path.join(ins_root, "instrument_{}_{}.wav".format(name,self.data["agg"])), self.mp.param["sr"], (np.array(wav_instrument) * 32768).astype("int16"), ) # @@ -155,7 +155,7 @@ class _audio_pre_: wav_vocals = spec_utils.cmb_spectrogram_to_wave(v_spec_m, self.mp) print("%s vocals done" % name) wavfile.write( - os.path.join(vocal_root, "vocal_{}.wav".format(name)), + os.path.join(vocal_root, "vocal_{}_{}.wav".format(name,self.data["agg"])), self.mp.param["sr"], (np.array(wav_vocals) * 32768).astype("int16"), ) diff --git a/train_nsf_sim_cache_sid_load_pretrain.py b/train_nsf_sim_cache_sid_load_pretrain.py index f7840f6..4ba6b65 100644 --- a/train_nsf_sim_cache_sid_load_pretrain.py +++ b/train_nsf_sim_cache_sid_load_pretrain.py @@ -45,7 +45,7 @@ global_step = 0 def main(): # n_gpus = torch.cuda.device_count() os.environ["MASTER_ADDR"] = "localhost" - os.environ["MASTER_PORT"] = "51515" + os.environ["MASTER_PORT"] = "51545" mp.spawn( run, diff --git a/vc_infer_pipeline.py b/vc_infer_pipeline.py index 13815e6..606812f 100644 --- a/vc_infer_pipeline.py +++ b/vc_infer_pipeline.py @@ -1,314 +1,313 @@ -import numpy as np, parselmouth, torch, pdb -from time import time as ttime -import torch.nn.functional as F -from config import x_pad, x_query, x_center, x_max -import scipy.signal as signal -import pyworld, os, traceback, faiss -from scipy import signal - -bh, ah = signal.butter(N=5, Wn=48, btype="high", fs=16000) - - -class VC(object): - def __init__(self, tgt_sr, device, is_half): - self.sr = 16000 # hubert输入采样率 - self.window = 160 # 每帧点数 - self.t_pad = self.sr * x_pad # 每条前后pad时间 - self.t_pad_tgt = tgt_sr * x_pad - self.t_pad2 = self.t_pad * 2 - self.t_query = self.sr * x_query # 查询切点前后查询时间 - self.t_center = self.sr * x_center # 查询切点位置 - self.t_max = self.sr * x_max # 免查询时长阈值 - self.device = device - self.is_half = is_half - - def get_f0(self, x, p_len, f0_up_key, f0_method, inp_f0=None): - time_step = self.window / self.sr * 1000 - f0_min = 50 - f0_max = 1100 - f0_mel_min = 1127 * np.log(1 + f0_min / 700) - f0_mel_max = 1127 * np.log(1 + f0_max / 700) - if f0_method == "pm": - f0 = ( - parselmouth.Sound(x, self.sr) - .to_pitch_ac( - time_step=time_step / 1000, - voicing_threshold=0.6, - pitch_floor=f0_min, - pitch_ceiling=f0_max, - ) - .selected_array["frequency"] - ) - pad_size = (p_len - len(f0) + 1) // 2 - if pad_size > 0 or p_len - len(f0) - pad_size > 0: - f0 = np.pad( - f0, [[pad_size, p_len - len(f0) - pad_size]], mode="constant" - ) - elif f0_method == "harvest": - f0, t = pyworld.harvest( - x.astype(np.double), - fs=self.sr, - f0_ceil=f0_max, - f0_floor=f0_min, - frame_period=10, - ) - f0 = pyworld.stonemask(x.astype(np.double), f0, t, self.sr) - f0 = signal.medfilt(f0, 3) - f0 *= pow(2, f0_up_key / 12) - # with open("test.txt","w")as f:f.write("\n".join([str(i)for i in f0.tolist()])) - tf0 = self.sr // self.window # 每秒f0点数 - if inp_f0 is not None: - delta_t = np.round( - (inp_f0[:, 0].max() - inp_f0[:, 0].min()) * tf0 + 1 - ).astype("int16") - replace_f0 = np.interp( - list(range(delta_t)), inp_f0[:, 0] * 100, inp_f0[:, 1] - ) - shape = f0[x_pad * tf0 : x_pad * tf0 + len(replace_f0)].shape[0] - f0[x_pad * tf0 : x_pad * tf0 + len(replace_f0)] = replace_f0[:shape] - # with open("test_opt.txt","w")as f:f.write("\n".join([str(i)for i in f0.tolist()])) - f0bak = f0.copy() - f0_mel = 1127 * np.log(1 + f0 / 700) - f0_mel[f0_mel > 0] = (f0_mel[f0_mel > 0] - f0_mel_min) * 254 / ( - f0_mel_max - f0_mel_min - ) + 1 - f0_mel[f0_mel <= 1] = 1 - f0_mel[f0_mel > 255] = 255 - f0_coarse = np.rint(f0_mel).astype(np.int) - return f0_coarse, f0bak # 1-0 - - def vc( - self, - model, - net_g, - sid, - audio0, - pitch, - pitchf, - times, - index, - big_npy, - index_rate, - ): # ,file_index,file_big_npy - feats = torch.from_numpy(audio0) - if self.is_half: - feats = feats.half() - else: - feats = feats.float() - if feats.dim() == 2: # double channels - feats = feats.mean(-1) - assert feats.dim() == 1, feats.dim() - feats = feats.view(1, -1) - padding_mask = torch.BoolTensor(feats.shape).to(self.device).fill_(False) - - inputs = { - "source": feats.to(self.device), - "padding_mask": padding_mask, - "output_layer": 9, # layer 9 - } - t0 = ttime() - with torch.no_grad(): - logits = model.extract_features(**inputs) - feats = model.final_proj(logits[0]) - - if ( - isinstance(index, type(None)) == False - and isinstance(big_npy, type(None)) == False - and index_rate != 0 - ): - npy = feats[0].cpu().numpy() - if self.is_half: - npy = npy.astype("float32") - - # _, I = index.search(npy, 1) - # npy = big_npy[I.squeeze()] - - #by github @nadare881 - score, ix = index.search(npy, k=8) - weight = np.square(1 / score) - weight /= weight.sum(axis=1, keepdims=True) - npy = np.sum(big_npy[ix] * np.expand_dims(weight, axis=2), axis=1) - - if self.is_half: - npy = npy.astype("float16") - feats = ( - torch.from_numpy(npy).unsqueeze(0).to(self.device) * index_rate - + (1 - index_rate) * feats - ) - - feats = F.interpolate(feats.permute(0, 2, 1), scale_factor=2).permute(0, 2, 1) - t1 = ttime() - p_len = audio0.shape[0] // self.window - if feats.shape[1] < p_len: - p_len = feats.shape[1] - if pitch != None and pitchf != None: - pitch = pitch[:, :p_len] - pitchf = pitchf[:, :p_len] - p_len = torch.tensor([p_len], device=self.device).long() - with torch.no_grad(): - if pitch != None and pitchf != None: - audio1 = ( - (net_g.infer(feats, p_len, pitch, pitchf, sid)[0][0, 0] * 32768) - .data.cpu() - .float() - .numpy() - .astype(np.int16) - ) - else: - audio1 = ( - (net_g.infer(feats, p_len, sid)[0][0, 0] * 32768) - .data.cpu() - .float() - .numpy() - .astype(np.int16) - ) - del feats, p_len, padding_mask - if torch.cuda.is_available(): - torch.cuda.empty_cache() - t2 = ttime() - times[0] += t1 - t0 - times[2] += t2 - t1 - return audio1 - - def pipeline( - self, - model, - net_g, - sid, - audio, - times, - f0_up_key, - f0_method, - file_index, - # file_big_npy, - index_rate, - if_f0, - f0_file=None, - ): - if ( - file_index != "" - # and file_big_npy != "" - # and os.path.exists(file_big_npy) == True - and os.path.exists(file_index) == True - and index_rate != 0 - ): - try: - index = faiss.read_index(file_index) - # big_npy = np.load(file_big_npy) - big_npy = index.reconstruct_n(0, index.ntotal) - except: - traceback.print_exc() - index = big_npy = None - else: - index = big_npy = None - audio = signal.filtfilt(bh, ah, audio) - audio_pad = np.pad(audio, (self.window // 2, self.window // 2), mode="reflect") - opt_ts = [] - if audio_pad.shape[0] > self.t_max: - audio_sum = np.zeros_like(audio) - for i in range(self.window): - audio_sum += audio_pad[i : i - self.window] - for t in range(self.t_center, audio.shape[0], self.t_center): - opt_ts.append( - t - - self.t_query - + np.where( - np.abs(audio_sum[t - self.t_query : t + self.t_query]) - == np.abs(audio_sum[t - self.t_query : t + self.t_query]).min() - )[0][0] - ) - s = 0 - audio_opt = [] - t = None - t1 = ttime() - audio_pad = np.pad(audio, (self.t_pad, self.t_pad), mode="reflect") - p_len = audio_pad.shape[0] // self.window - inp_f0 = None - if hasattr(f0_file, "name") == True: - try: - with open(f0_file.name, "r") as f: - lines = f.read().strip("\n").split("\n") - inp_f0 = [] - for line in lines: - inp_f0.append([float(i) for i in line.split(",")]) - inp_f0 = np.array(inp_f0, dtype="float32") - except: - traceback.print_exc() - sid = torch.tensor(sid, device=self.device).unsqueeze(0).long() - pitch, pitchf = None, None - if if_f0 == 1: - pitch, pitchf = self.get_f0(audio_pad, p_len, f0_up_key, f0_method, inp_f0) - pitch = pitch[:p_len] - pitchf = pitchf[:p_len] - pitch = torch.tensor(pitch, device=self.device).unsqueeze(0).long() - pitchf = torch.tensor(pitchf, device=self.device).unsqueeze(0).float() - t2 = ttime() - times[1] += t2 - t1 - for t in opt_ts: - t = t // self.window * self.window - if if_f0 == 1: - audio_opt.append( - self.vc( - model, - net_g, - sid, - audio_pad[s : t + self.t_pad2 + self.window], - pitch[:, s // self.window : (t + self.t_pad2) // self.window], - pitchf[:, s // self.window : (t + self.t_pad2) // self.window], - times, - index, - big_npy, - index_rate, - )[self.t_pad_tgt : -self.t_pad_tgt] - ) - else: - audio_opt.append( - self.vc( - model, - net_g, - sid, - audio_pad[s : t + self.t_pad2 + self.window], - None, - None, - times, - index, - big_npy, - index_rate, - )[self.t_pad_tgt : -self.t_pad_tgt] - ) - s = t - if if_f0 == 1: - audio_opt.append( - self.vc( - model, - net_g, - sid, - audio_pad[t:], - pitch[:, t // self.window :] if t is not None else pitch, - pitchf[:, t // self.window :] if t is not None else pitchf, - times, - index, - big_npy, - index_rate, - )[self.t_pad_tgt : -self.t_pad_tgt] - ) - else: - audio_opt.append( - self.vc( - model, - net_g, - sid, - audio_pad[t:], - None, - None, - times, - index, - big_npy, - index_rate, - )[self.t_pad_tgt : -self.t_pad_tgt] - ) - audio_opt = np.concatenate(audio_opt) - del pitch, pitchf, sid - if torch.cuda.is_available(): - torch.cuda.empty_cache() - return audio_opt +import numpy as np, parselmouth, torch, pdb +from time import time as ttime +import torch.nn.functional as F +from config import x_pad, x_query, x_center, x_max +import scipy.signal as signal +import pyworld, os, traceback, faiss +from scipy import signal + +bh, ah = signal.butter(N=5, Wn=48, btype="high", fs=16000) + + +class VC(object): + def __init__(self, tgt_sr, device, is_half): + self.sr = 16000 # hubert输入采样率 + self.window = 160 # 每帧点数 + self.t_pad = self.sr * x_pad # 每条前后pad时间 + self.t_pad_tgt = tgt_sr * x_pad + self.t_pad2 = self.t_pad * 2 + self.t_query = self.sr * x_query # 查询切点前后查询时间 + self.t_center = self.sr * x_center # 查询切点位置 + self.t_max = self.sr * x_max # 免查询时长阈值 + self.device = device + self.is_half = is_half + + def get_f0(self, x, p_len, f0_up_key, f0_method, inp_f0=None): + time_step = self.window / self.sr * 1000 + f0_min = 50 + f0_max = 1100 + f0_mel_min = 1127 * np.log(1 + f0_min / 700) + f0_mel_max = 1127 * np.log(1 + f0_max / 700) + if f0_method == "pm": + f0 = ( + parselmouth.Sound(x, self.sr) + .to_pitch_ac( + time_step=time_step / 1000, + voicing_threshold=0.6, + pitch_floor=f0_min, + pitch_ceiling=f0_max, + ) + .selected_array["frequency"] + ) + pad_size = (p_len - len(f0) + 1) // 2 + if pad_size > 0 or p_len - len(f0) - pad_size > 0: + f0 = np.pad( + f0, [[pad_size, p_len - len(f0) - pad_size]], mode="constant" + ) + elif f0_method == "harvest": + f0, t = pyworld.harvest( + x.astype(np.double), + fs=self.sr, + f0_ceil=f0_max, + f0_floor=f0_min, + frame_period=10, + ) + f0 = pyworld.stonemask(x.astype(np.double), f0, t, self.sr) + f0 = signal.medfilt(f0, 3) + f0 *= pow(2, f0_up_key / 12) + # with open("test.txt","w")as f:f.write("\n".join([str(i)for i in f0.tolist()])) + tf0 = self.sr // self.window # 每秒f0点数 + if inp_f0 is not None: + delta_t = np.round( + (inp_f0[:, 0].max() - inp_f0[:, 0].min()) * tf0 + 1 + ).astype("int16") + replace_f0 = np.interp( + list(range(delta_t)), inp_f0[:, 0] * 100, inp_f0[:, 1] + ) + shape = f0[x_pad * tf0 : x_pad * tf0 + len(replace_f0)].shape[0] + f0[x_pad * tf0 : x_pad * tf0 + len(replace_f0)] = replace_f0[:shape] + # with open("test_opt.txt","w")as f:f.write("\n".join([str(i)for i in f0.tolist()])) + f0bak = f0.copy() + f0_mel = 1127 * np.log(1 + f0 / 700) + f0_mel[f0_mel > 0] = (f0_mel[f0_mel > 0] - f0_mel_min) * 254 / ( + f0_mel_max - f0_mel_min + ) + 1 + f0_mel[f0_mel <= 1] = 1 + f0_mel[f0_mel > 255] = 255 + f0_coarse = np.rint(f0_mel).astype(np.int) + return f0_coarse, f0bak # 1-0 + + def vc( + self, + model, + net_g, + sid, + audio0, + pitch, + pitchf, + times, + index, + big_npy, + index_rate, + ): # ,file_index,file_big_npy + feats = torch.from_numpy(audio0) + if self.is_half: + feats = feats.half() + else: + feats = feats.float() + if feats.dim() == 2: # double channels + feats = feats.mean(-1) + assert feats.dim() == 1, feats.dim() + feats = feats.view(1, -1) + padding_mask = torch.BoolTensor(feats.shape).to(self.device).fill_(False) + + inputs = { + "source": feats.to(self.device), + "padding_mask": padding_mask, + "output_layer": 9, # layer 9 + } + t0 = ttime() + with torch.no_grad(): + logits = model.extract_features(**inputs) + feats = model.final_proj(logits[0]) + + if ( + isinstance(index, type(None)) == False + and isinstance(big_npy, type(None)) == False + and index_rate != 0 + ): + npy = feats[0].cpu().numpy() + if self.is_half: + npy = npy.astype("float32") + + # _, I = index.search(npy, 1) + # npy = big_npy[I.squeeze()] + + score, ix = index.search(npy, k=8) + weight = np.square(1 / score) + weight /= weight.sum(axis=1, keepdims=True) + npy = np.sum(big_npy[ix] * np.expand_dims(weight, axis=2), axis=1) + + if self.is_half: + npy = npy.astype("float16") + feats = ( + torch.from_numpy(npy).unsqueeze(0).to(self.device) * index_rate + + (1 - index_rate) * feats + ) + + feats = F.interpolate(feats.permute(0, 2, 1), scale_factor=2).permute(0, 2, 1) + t1 = ttime() + p_len = audio0.shape[0] // self.window + if feats.shape[1] < p_len: + p_len = feats.shape[1] + if pitch != None and pitchf != None: + pitch = pitch[:, :p_len] + pitchf = pitchf[:, :p_len] + p_len = torch.tensor([p_len], device=self.device).long() + with torch.no_grad(): + if pitch != None and pitchf != None: + audio1 = ( + (net_g.infer(feats, p_len, pitch, pitchf, sid)[0][0, 0] * 32768) + .data.cpu() + .float() + .numpy() + .astype(np.int16) + ) + else: + audio1 = ( + (net_g.infer(feats, p_len, sid)[0][0, 0] * 32768) + .data.cpu() + .float() + .numpy() + .astype(np.int16) + ) + del feats, p_len, padding_mask + if torch.cuda.is_available(): + torch.cuda.empty_cache() + t2 = ttime() + times[0] += t1 - t0 + times[2] += t2 - t1 + return audio1 + + def pipeline( + self, + model, + net_g, + sid, + audio, + times, + f0_up_key, + f0_method, + file_index, + # file_big_npy, + index_rate, + if_f0, + f0_file=None, + ): + if ( + file_index != "" + # and file_big_npy != "" + # and os.path.exists(file_big_npy) == True + and os.path.exists(file_index) == True + and index_rate != 0 + ): + try: + index = faiss.read_index(file_index) + # big_npy = np.load(file_big_npy) + big_npy = index.reconstruct_n(0, index.ntotal) + except: + traceback.print_exc() + index = big_npy = None + else: + index = big_npy = None + audio = signal.filtfilt(bh, ah, audio) + audio_pad = np.pad(audio, (self.window // 2, self.window // 2), mode="reflect") + opt_ts = [] + if audio_pad.shape[0] > self.t_max: + audio_sum = np.zeros_like(audio) + for i in range(self.window): + audio_sum += audio_pad[i : i - self.window] + for t in range(self.t_center, audio.shape[0], self.t_center): + opt_ts.append( + t + - self.t_query + + np.where( + np.abs(audio_sum[t - self.t_query : t + self.t_query]) + == np.abs(audio_sum[t - self.t_query : t + self.t_query]).min() + )[0][0] + ) + s = 0 + audio_opt = [] + t = None + t1 = ttime() + audio_pad = np.pad(audio, (self.t_pad, self.t_pad), mode="reflect") + p_len = audio_pad.shape[0] // self.window + inp_f0 = None + if hasattr(f0_file, "name") == True: + try: + with open(f0_file.name, "r") as f: + lines = f.read().strip("\n").split("\n") + inp_f0 = [] + for line in lines: + inp_f0.append([float(i) for i in line.split(",")]) + inp_f0 = np.array(inp_f0, dtype="float32") + except: + traceback.print_exc() + sid = torch.tensor(sid, device=self.device).unsqueeze(0).long() + pitch, pitchf = None, None + if if_f0 == 1: + pitch, pitchf = self.get_f0(audio_pad, p_len, f0_up_key, f0_method, inp_f0) + pitch = pitch[:p_len] + pitchf = pitchf[:p_len] + pitch = torch.tensor(pitch, device=self.device).unsqueeze(0).long() + pitchf = torch.tensor(pitchf, device=self.device).unsqueeze(0).float() + t2 = ttime() + times[1] += t2 - t1 + for t in opt_ts: + t = t // self.window * self.window + if if_f0 == 1: + audio_opt.append( + self.vc( + model, + net_g, + sid, + audio_pad[s : t + self.t_pad2 + self.window], + pitch[:, s // self.window : (t + self.t_pad2) // self.window], + pitchf[:, s // self.window : (t + self.t_pad2) // self.window], + times, + index, + big_npy, + index_rate, + )[self.t_pad_tgt : -self.t_pad_tgt] + ) + else: + audio_opt.append( + self.vc( + model, + net_g, + sid, + audio_pad[s : t + self.t_pad2 + self.window], + None, + None, + times, + index, + big_npy, + index_rate, + )[self.t_pad_tgt : -self.t_pad_tgt] + ) + s = t + if if_f0 == 1: + audio_opt.append( + self.vc( + model, + net_g, + sid, + audio_pad[t:], + pitch[:, t // self.window :] if t is not None else pitch, + pitchf[:, t // self.window :] if t is not None else pitchf, + times, + index, + big_npy, + index_rate, + )[self.t_pad_tgt : -self.t_pad_tgt] + ) + else: + audio_opt.append( + self.vc( + model, + net_g, + sid, + audio_pad[t:], + None, + None, + times, + index, + big_npy, + index_rate, + )[self.t_pad_tgt : -self.t_pad_tgt] + ) + audio_opt = np.concatenate(audio_opt) + del pitch, pitchf, sid + if torch.cuda.is_available(): + torch.cuda.empty_cache() + return audio_opt