GPT-SoVITS

model arch

• S1 model: AR model–ssl tokens
• S2 model: VITS，ssl 已经是mel 长度线性相关，MRTE(ssl_codes_embs, text, global_mel_emb)模块，将文本加强相关，学到一个参考结果

S1 Model

class Text2SemanticDecoder()def forward_old(self, x, x_lens, y, y_lens, bert_feature):"""x: phoneme_idsy: semantic_idsbert_feature: 已经根据word2phn 扩展成和x等长train : y+EOS，已知长度；infer : AR 预测，预测EOS 终止；如果没有，到预设最大长度，终止；"""# phn torch.Size([20, 99]) bert_feature torch.Size([20, 1024, 99])x = self.ar_text_embedding(x)x = x + self.bert_proj(bert_feature.transpose(1, 2))x = self.ar_text_position(x)x_mask = make_pad_mask(x_lens)y_mask = make_pad_mask(y_lens)y_mask_int = y_mask.type(torch.int64)codes = y.type(torch.int64) * (1 - y_mask_int)# Training# AR Decoder： SinePositionalEmbeddingy, targets = self.pad_y_eos(codes, y_mask_int, eos_id=self.EOS)x_len = x_lens.max()y_len = y_lens.max()y_emb = self.ar_audio_embedding(y)y_pos = self.ar_audio_position(y_emb)xy_padding_mask = torch.concat([x_mask, y_mask], dim=1)ar_xy_padding_mask = xy_padding_maskx_attn_mask = F.pad(torch.zeros((x_len, x_len), dtype=torch.bool, device=x.device),(0, y_len),value=True,)y_attn_mask = F.pad(torch.triu(torch.ones(y_len, y_len, dtype=torch.bool, device=x.device),diagonal=1,),(x_len, 0),value=False,)xy_attn_mask = torch.concat([x_attn_mask, y_attn_mask], dim=0)bsz, src_len = x.shape[0], x_len + y_len_xy_padding_mask = (ar_xy_padding_mask.view(bsz, 1, 1, src_len).expand(-1, self.num_head, -1, -1).reshape(bsz * self.num_head, 1, src_len))xy_attn_mask = xy_attn_mask.logical_or(_xy_padding_mask)new_attn_mask = torch.zeros_like(xy_attn_mask, dtype=x.dtype)new_attn_mask.masked_fill_(xy_attn_mask, float("-inf"))xy_attn_mask = new_attn_mask# x 和完整的 y 一次性输入模型xy_pos = torch.concat([x, y_pos], dim=1)xy_dec, _ = self.h((xy_pos, None),mask=xy_attn_mask,)logits = self.ar_predict_layer(xy_dec[:, x_len:]).permute(0, 2, 1)# loss# from feiteng: 每次 duration 越多, 梯度更新也应该更多, 所以用 sumloss = F.cross_entropy(logits, targets, reduction="sum")acc = self.ar_accuracy_metric(logits.detach(), targets).item()return loss, acc

S2 model

class Encoder()def forward(self, ssl, y_lengths, text, text_lengths, speed=1,test=None):'''y_lengths: mel_lengthge : ref_encoder_outputs'''ge = self.ref_enc(y * y_mask, y_mask)ssl = self.ssl_proj(ssl)quantized, codes, commit_loss, quantized_list = self.quantizer(ssl, layers=[0])if self.semantic_frame_rate == "25hz":quantized = F.interpolate(quantized, size=int(quantized.shape[-1] * 2), mode="nearest")y = self.encoder_ssl(y * y_mask, y_mask)text_mask = torch.unsqueeze(commons.sequence_mask(text_lengths, text.size(1)), 1).to(y.dtype)if test == 1:text[:, :] = 0text = self.text_embedding(text).transpose(1, 2)text = self.encoder_text(text * text_mask, text_mask)y = self.mrte(y, y_mask, text, text_mask, ge)# self.encoder_ssl, self.encoder_text, self.encoder2 结构一样y = self.encoder2(y * y_mask, y_mask)if(speed!=1):y = F.interpolate(y, size=int(y.shape[-1] / speed)+1, mode="linear")y_mask = F.interpolate(y_mask, size=y.shape[-1], mode="nearest")stats = self.proj(y) * y_maskm, logs = torch.split(stats, self.out_channels, dim=1)return y, m, logs, y_mask