search

Artrajz / vits-simple-api

A simple VITS HTTP API, developed by extending Moegoe with additional features.
GNU Affero General Public License v3.0
777 stars 116 forks source link

bert-vits2 语言无法正确选择 #172

Closed grider-withourai closed 1 month ago

grider-withourai commented 1 month ago

运行环境

问题描述

已更改模型配置文件为

  "data": {
    "lang": ["ja"],
    "training_files": "filelists/train.list",
    "validation_files": "filelists/val.list",
...

但模型加载还是只加载了中文bert

INFO:root:Loading config...
INFO:root:Loading config success!
INFO:root:Config is saved.
Building prefix dict from the default dictionary ...
DEBUG:jieba:Building prefix dict from the default dictionary ...
Dumping model to file cache /tmp/jieba.cache
DEBUG:jieba:Dumping model to file cache /tmp/jieba.cache
Loading model cost 1.211 seconds.
DEBUG:jieba:Loading model cost 1.211 seconds.
Prefix dict has been built successfully.
DEBUG:jieba:Prefix dict has been built successfully.
INFO:langid.langid:initializing identifier
2024-07-18 09:16:22 [INFO] Loading BERT model: /app/data/bert/Erlangshen-MegatronBert-1.3B-Chinese [in model_handler.load_bert:168]
2024-07-18 09:16:35 [INFO] Success loading: /app/data/bert/Erlangshen-MegatronBert-1.3B-Chinese [in model_handler.load_bert:179]
2024-07-18 09:16:35 [INFO] Loading CLAP_HTSAT_FUSED: /app/data/emotional/clap-htsat-fused [in model_handler.load_clap:237]
2024-07-18 09:16:40 [INFO] Success loading: /app/data/emotional/clap-htsat-fused [in model_handler.load_clap:244]
2024-07-18 09:16:40 [INFO] Loading G2PWModel: /app/data/G2PWModel [in model_handler.get_pinyinPlus:282]
/usr/local/lib/python3.10/site-packages/onnxruntime/capi/onnxruntime_inference_collection.py:69: UserWarning: Specified provider 'CUDAExecutionProvider' is not in available provider names.Available providers: 'AzureExecutionProvider, CPUExecutionProvider'
  warnings.warn(
2024-07-18 09:16:44 [INFO] Success loading G2PWModel [in model_handler.get_pinyinPlus:289]
2024-07-18 09:16:47 [ERROR] enc_p.bert_pre_proj.weight is not in the checkpoint [in utils.load_checkpoint:59]
2024-07-18 09:16:47 [ERROR] enc_p.bert_pre_proj.bias is not in the checkpoint [in utils.load_checkpoint:59]
2024-07-18 09:16:47 [ERROR] enc_p.emo_vq.project_in.weight is not in the checkpoint [in utils.load_checkpoint:59]
2024-07-18 09:16:47 [ERROR] enc_p.emo_vq.project_in.bias is not in the checkpoint [in utils.load_checkpoint:59]
2024-07-18 09:16:47 [ERROR] enc_p.emo_vq.project_out.weight is not in the checkpoint [in utils.load_checkpoint:59]
2024-07-18 09:16:47 [ERROR] enc_p.emo_vq.project_out.bias is not in the checkpoint [in utils.load_checkpoint:59]
2024-07-18 09:16:47 [ERROR] enc_p.emo_vq._codebook.cluster_size is not in the checkpoint [in utils.load_checkpoint:59]
2024-07-18 09:16:47 [ERROR] enc_p.emo_vq._codebook.embed_avg is not in the checkpoint [in utils.load_checkpoint:59]
2024-07-18 09:16:47 [ERROR] enc_p.emo_vq._codebook.embed is not in the checkpoint [in utils.load_checkpoint:59]
2024-07-18 09:16:47 [INFO] Loaded checkpoint '/app/data/models/chocola/G_7000.pth' (iteration 467) [in utils.load_checkpoint:67]
2024-07-18 09:16:47 [INFO] model_type:BERT-VITS2 model_id:0 n_speakers:1 model_path:/app/data/models/chocola/G_7000.pth [in ModelManager._load_model_from_path:310]
2024-07-18 09:16:47 [INFO] PyTorch Version: 2.2.1+cu118 Cuda available:True Device type:cuda [in ModelManager.log_device_info:164]
2024-07-18 09:16:47 [INFO] Using GPU on NVIDIA GeForce RTX 4090 24GB, GPU Device Index: None [in ModelManager.log_device_info:170]
2024-07-18 09:16:47 [INFO] [BERT-VITS2] 1 speakers [in ModelManager.model_init:110]
2024-07-18 09:16:47 [INFO] 1 speakers in total. [in ModelManager.model_init:113]
2024-07-18 09:16:47 [INFO] Loading phrases_dict [in phrases_dict.phrases_dict_init:40]
2024-07-18 09:16:48 [INFO] Additional phrases loaded from /app/data/phrases_dict.txt [in phrases_dict.load_phrases_from_file:27]
2024-07-18 09:16:48 [INFO] Scheduler started [in base.start:181]
2024-07-18 09:16:48 [INFO] Added job "clean_task" to job store "default" [in base._real_add_job:895]
[2024-07-18 09:16:48 +0800] [1] [INFO] Starting gunicorn 22.0.0
[2024-07-18 09:16:48 +0800] [1] [INFO] Listening at: http://0.0.0.0:23456 (1)
[2024-07-18 09:16:48 +0800] [1] [INFO] Using worker: sync
[2024-07-18 09:16:48 +0800] [138] [INFO] Booting worker with pid: 138

且前端显示中文zh image

grider-withourai commented 1 month ago

查阅源码发现日文特化版本需要把version设为ja_extra。