f197177ae1
- FastAPI 백엔드: /api/voices, /api/tts, /api/health - 한국어 메인 + 영어(5종 악센트), 중국어/일본어 UI 제거 - 언어별 목소리 모델 선택, 속도(0.5~2.0x)/피치(±12반음) 조절 - 로그인/사용제한/과금 유도 없음 (무제한 테스트) - NVIDIA GPU 자동 사용 (torch cu128, Blackwell sm_120) - 유저 친화 한국어 UI (frontend) - Docker/compose (GPU 예약), 8788 포트 - MeloTTS(순수 파이썬) 벤더링 + 검증된 버전 고정(constraints)
108 lines
2.5 KiB
Python
108 lines
2.5 KiB
Python
import torch
|
|
import sys
|
|
from transformers import AutoTokenizer, AutoModelForMaskedLM
|
|
|
|
|
|
# model_id = 'hfl/chinese-roberta-wwm-ext-large'
|
|
local_path = "./bert/chinese-roberta-wwm-ext-large"
|
|
|
|
|
|
tokenizers = {}
|
|
models = {}
|
|
|
|
def get_bert_feature(text, word2ph, device=None, model_id='hfl/chinese-roberta-wwm-ext-large'):
|
|
if model_id not in models:
|
|
models[model_id] = AutoModelForMaskedLM.from_pretrained(
|
|
model_id
|
|
).to(device)
|
|
tokenizers[model_id] = AutoTokenizer.from_pretrained(model_id)
|
|
model = models[model_id]
|
|
tokenizer = tokenizers[model_id]
|
|
|
|
if (
|
|
sys.platform == "darwin"
|
|
and torch.backends.mps.is_available()
|
|
and device == "cpu"
|
|
):
|
|
device = "mps"
|
|
if not device:
|
|
device = "cuda"
|
|
|
|
with torch.no_grad():
|
|
inputs = tokenizer(text, return_tensors="pt")
|
|
for i in inputs:
|
|
inputs[i] = inputs[i].to(device)
|
|
res = model(**inputs, output_hidden_states=True)
|
|
res = torch.cat(res["hidden_states"][-3:-2], -1)[0].cpu()
|
|
# import pdb; pdb.set_trace()
|
|
# assert len(word2ph) == len(text) + 2
|
|
word2phone = word2ph
|
|
phone_level_feature = []
|
|
for i in range(len(word2phone)):
|
|
repeat_feature = res[i].repeat(word2phone[i], 1)
|
|
phone_level_feature.append(repeat_feature)
|
|
|
|
phone_level_feature = torch.cat(phone_level_feature, dim=0)
|
|
return phone_level_feature.T
|
|
|
|
|
|
if __name__ == "__main__":
|
|
import torch
|
|
|
|
word_level_feature = torch.rand(38, 1024) # 12个词,每个词1024维特征
|
|
word2phone = [
|
|
1,
|
|
2,
|
|
1,
|
|
2,
|
|
2,
|
|
1,
|
|
2,
|
|
2,
|
|
1,
|
|
2,
|
|
2,
|
|
1,
|
|
2,
|
|
2,
|
|
2,
|
|
2,
|
|
2,
|
|
1,
|
|
1,
|
|
2,
|
|
2,
|
|
1,
|
|
2,
|
|
2,
|
|
2,
|
|
2,
|
|
1,
|
|
2,
|
|
2,
|
|
2,
|
|
2,
|
|
2,
|
|
1,
|
|
2,
|
|
2,
|
|
2,
|
|
2,
|
|
1,
|
|
]
|
|
|
|
# 计算总帧数
|
|
total_frames = sum(word2phone)
|
|
print(word_level_feature.shape)
|
|
print(word2phone)
|
|
phone_level_feature = []
|
|
for i in range(len(word2phone)):
|
|
print(word_level_feature[i].shape)
|
|
|
|
# 对每个词重复word2phone[i]次
|
|
repeat_feature = word_level_feature[i].repeat(word2phone[i], 1)
|
|
phone_level_feature.append(repeat_feature)
|
|
|
|
phone_level_feature = torch.cat(phone_level_feature, dim=0)
|
|
print(phone_level_feature.shape) # torch.Size([36, 1024])
|