import os import numpy as np import librosa,torch from feature_extractor import cnhubert from module.models import SynthesizerTrn from AR.models.t2s_lightning_module import Text2SemanticLightningModule from transformers import AutoModelForMaskedLM, AutoTokenizer from text import cleaned_text_to_sequence from text.cleaner import clean_text from time import time as ttime from module.mel_processing import spectrogram_torch from my_utils import load_audio class DictToAttrRecursive(dict): def __init__(self, input_dict): super().__init__(input_dict) for key, value in input_dict.items(): if isinstance(value, dict): value = DictToAttrRecursive(value) self[key] = value setattr(self, key, value) def __getattr__(self, item): try: return self[item] except KeyError: raise AttributeError(f"Attribute {item} not found") def __setattr__(self, key, value): if isinstance(value, dict): value = DictToAttrRecursive(value) super(DictToAttrRecursive, self).__setitem__(key, value) super().__setattr__(key, value) def __delattr__(self, item): try: del self[item] except KeyError: raise AttributeError(f"Attribute {item} not found") class Inference: def __init__(self, is_half, GPT_weight_root, SoVITS_weight_root): self.n_semantic = 1024 self.model_loaded = False self.is_half = is_half self.GPT_weight_root = GPT_weight_root self.SoVITS_weight_root = SoVITS_weight_root def update_envs(self, gpt_path, sovits_path, cnhubert_base_path, bert_path): self.gpt_path = os.path.join(self.GPT_weight_root, gpt_path) self.sovits_path = os.path.join(self.SoVITS_weight_root, sovits_path) self.cnhubert_base_path = cnhubert_base_path self.bert_path = bert_path cnhubert.cnhubert_base_path=cnhubert_base_path yield self.load_model() def load_model(self, device='cuda'): try: # Load bert model self.device = device self.tokenizer = AutoTokenizer.from_pretrained(self.bert_path) self.bert_model = AutoModelForMaskedLM.from_pretrained(self.bert_path) if self.is_half == True: self.bert_model = self.bert_model.half().to(device) else: self.bert_model = self.bert_model.to(device) # Load ssl model dict_s1 = torch.load(self.gpt_path, map_location="cpu") self.config = dict_s1["config"] self.ssl_model = cnhubert.get_model() if self.is_half == True: self.ssl_model = self.ssl_model.half().to(device) else: self.ssl_model = self.ssl_model.to(device) dict_s2=torch.load(self.sovits_path,map_location="cpu") self.hps=dict_s2["config"] self.hps = DictToAttrRecursive(self.hps) self.hps.model.semantic_frame_rate = "25hz" # Load vq model self.vq_model = SynthesizerTrn( self.hps.data.filter_length // 2 + 1, self.hps.train.segment_size // self.hps.data.hop_length, n_speakers=self.hps.data.n_speakers, **self.hps.model ) if self.is_half == True: self.vq_model = self.vq_model.half().to(device) else: self.vq_model = self.vq_model.to(device) self.vq_model.eval() self.vq_model.load_state_dict(dict_s2["weight"], strict=False) # Load t2s model # t2s_model = Text2SemanticLightningModule.load_from_checkpoint(checkpoint_path=gpt_path, config=config, map_location="cpu")#########todo self.t2s_model = Text2SemanticLightningModule(self.config, "ojbk", is_train=False) self.t2s_model.load_state_dict(dict_s1["weight"]) if self.is_half == True: self.t2s_model = self.t2s_model.half() self.t2s_model = self.t2s_model.to(device) self.t2s_model.eval() total = sum([param.nelement() for param in self.t2s_model.parameters()]) print("Number of parameter: %.2fM" % (total / 1e6)) self.model_loaded = True return '模型加载成功' except Exception as e: return f'模型加载失败：{e}' def unload_model(self): if self.model_loaded: try: del self.bert_model, self.ssl_model, self.hps, self.vq_model, self.t2s_model if torch.cuda.is_available(): torch.cuda.empty_cache() self.model_loaded = False yield '模型卸载成功' except Exception as e: yield f'模型卸载失败：{e}' else: yield '模型未加载' def get_bert_feature(self, text, word2ph): with torch.no_grad(): inputs = self.tokenizer(text, return_tensors="pt") for i in inputs: inputs[i] = inputs[i].to(self.device) #####输入是long不用管精度问题，精度随bert_model res = self.bert_model(**inputs, output_hidden_states=True) res = torch.cat(res["hidden_states"][-3:-2], -1)[0].cpu()[1:-1] assert len(word2ph) == len(text) phone_level_feature = [] for i in range(len(word2ph)): repeat_feature = res[i].repeat(word2ph[i], 1) phone_level_feature.append(repeat_feature) phone_level_feature = torch.cat(phone_level_feature, dim=0) # if(is_half==True):phone_level_feature=phone_level_feature.half() return phone_level_feature.T def get_tts_wav(self, ref_wav_path, prompt_text, prompt_language, text, text_language): if not self.model_loaded: return hz = 50 dict_language = {"中文": "zh", "英文": "en", "日文": "ja"} t0 = ttime() prompt_text = prompt_text.strip("\n") prompt_language, text = prompt_language, text.strip("\n") with torch.no_grad(): wav16k, sr = librosa.load(ref_wav_path, sr=16000) # 派蒙 wav16k = torch.from_numpy(wav16k) if self.is_half == True: wav16k = wav16k.half().to(self.device) else: wav16k = wav16k.to(self.device) ssl_content = self.ssl_model.model(wav16k.unsqueeze(0))[ "last_hidden_state" ].transpose( 1, 2 ) # .float() codes = self.vq_model.extract_latent(ssl_content) prompt_semantic = codes[0, 0] t1 = ttime() prompt_language = dict_language[prompt_language] text_language = dict_language[text_language] phones1, word2ph1, norm_text1 = clean_text(prompt_text, prompt_language) phones1 = cleaned_text_to_sequence(phones1) texts = text.split("\n") audio_opt = [] zero_wav = np.zeros( int(self.hps.data.sampling_rate * 0.3), dtype=np.float16 if self.is_half == True else np.float32, ) for text in texts: # 解决输入目标文本的空行导致报错的问题 if (len(text.strip()) == 0): continue phones2, word2ph2, norm_text2 = clean_text(text, text_language) phones2 = cleaned_text_to_sequence(phones2) if prompt_language == "zh": bert1 = self.get_bert_feature(norm_text1, word2ph1).to(self.device) else: bert1 = torch.zeros( (1024, len(phones1)), dtype=torch.float16 if self.is_half == True else torch.float32, ).to(self.device) if text_language == "zh": bert2 = self.get_bert_feature(norm_text2, word2ph2).to(self.device) else: bert2 = torch.zeros((1024, len(phones2))).to(bert1) bert = torch.cat([bert1, bert2], 1) all_phoneme_ids = torch.LongTensor(phones1 + phones2).to(self.device).unsqueeze(0) bert = bert.to(self.device).unsqueeze(0) all_phoneme_len = torch.tensor([all_phoneme_ids.shape[-1]]).to(self.device) prompt = prompt_semantic.unsqueeze(0).to(self.device) t2 = ttime() with torch.no_grad(): # pred_semantic = t2s_model.model.infer( pred_semantic, idx = self.t2s_model.model.infer_panel( all_phoneme_ids, all_phoneme_len, prompt, bert, # prompt_phone_len=ph_offset, top_k=self.config["inference"]["top_k"], early_stop_num=hz * self.config["data"]["max_sec"] ) t3 = ttime() # print(pred_semantic.shape,idx) pred_semantic = pred_semantic[:, -idx:].unsqueeze( 0 ) # .unsqueeze(0)#mq要多unsqueeze一次 refer = get_spepc(self.hps, ref_wav_path) # .to(device) if self.is_half == True: refer = refer.half().to(self.device) else: refer = refer.to(self.device) # audio = vq_model.decode(pred_semantic, all_phoneme_ids, refer).detach().cpu().numpy()[0, 0] audio = ( self.vq_model.decode( pred_semantic, torch.LongTensor(phones2).to(self.device).unsqueeze(0), refer ) .detach() .cpu() .numpy()[0, 0] ) ###试试重建不带上prompt部分 audio_opt.append(audio) audio_opt.append(zero_wav) t4 = ttime() print("%.3f\t%.3f\t%.3f\t%.3f" % (t1 - t0, t2 - t1, t3 - t2, t4 - t3)) yield self.hps.data.sampling_rate, (np.concatenate(audio_opt, 0) * 32768).astype( np.int16 ) def get_spepc(hps, filename): audio = load_audio(filename, int(hps.data.sampling_rate)) audio = torch.FloatTensor(audio) audio_norm = audio audio_norm = audio_norm.unsqueeze(0) spec = spectrogram_torch( audio_norm, hps.data.filter_length, hps.data.sampling_rate, hps.data.hop_length, hps.data.win_length, center=False, ) return spec def split(todo_text): splits = { "，", "。", "？", "！", ",", ".", "?", "!", "~", ":", "：", "—", "…", } # 不考虑省略号 todo_text = todo_text.replace("……", "。").replace("——", "，") if todo_text[-1] not in splits: todo_text += "。" i_split_head = i_split_tail = 0 len_text = len(todo_text) todo_texts = [] while 1: if i_split_head >= len_text: break # 结尾一定有标点，所以直接跳出即可，最后一段在上次已加入 if todo_text[i_split_head] in splits: i_split_head += 1 todo_texts.append(todo_text[i_split_tail:i_split_head]) i_split_tail = i_split_head else: i_split_head += 1 return todo_texts def cut1(inp): inp = inp.strip("\n") inps = split(inp) split_idx = list(range(0, len(inps), 5)) split_idx[-1] = None if len(split_idx) > 1: opts = [] for idx in range(len(split_idx) - 1): opts.append("".join(inps[split_idx[idx] : split_idx[idx + 1]])) else: opts = [inp] return "\n".join(opts) def cut2(inp): inp = inp.strip("\n") inps = split(inp) if len(inps) < 2: return [inp] opts = [] summ = 0 tmp_str = "" for i in range(len(inps)): summ += len(inps[i]) tmp_str += inps[i] if summ > 50: summ = 0 opts.append(tmp_str) tmp_str = "" if tmp_str != "": opts.append(tmp_str) if len(opts[-1]) < 50: ##如果最后一个太短了，和前一个合一起 opts[-2] = opts[-2] + opts[-1] opts = opts[:-1] return "\n".join(opts) def cut3(inp): inp = inp.strip("\n") return "\n".join(["%s。" % item for item in inp.strip("。").split("。")])