Merge af364358bedea07d7851613bbe9fef2a1b72bb40 into 836bfec1fbf7356d59bd2dbe3883997646554e20

GPT_SoVITS/inference_webui.py

@@ -25,11 +25,9 @@ import torch
 
 infer_ttswebui = os.environ.get("infer_ttswebui", 9872)
 infer_ttswebui = int(infer_ttswebui)
-is_share = os.environ.get("is_share", "False")
-is_share = eval(is_share)
 if "_CUDA_VISIBLE_DEVICES" in os.environ:
     os.environ["CUDA_VISIBLE_DEVICES"] = os.environ["_CUDA_VISIBLE_DEVICES"]
-is_half = eval(os.environ.get("is_half", "True")) and torch.cuda.is_available()
+from config import is_half,is_share
 gpt_path = os.environ.get("gpt_path", None)
 sovits_path = os.environ.get("sovits_path", None)
 cnhubert_base_path = os.environ.get("cnhubert_base_path", None)

GPT_SoVITS/prepare_datasets/1-get-text.py

@@ -10,7 +10,7 @@ all_parts = os.environ.get("all_parts")
 os.environ["CUDA_VISIBLE_DEVICES"] = os.environ.get("_CUDA_VISIBLE_DEVICES")
 opt_dir = os.environ.get("opt_dir")
 bert_pretrained_dir = os.environ.get("bert_pretrained_dir")
-is_half = eval(os.environ.get("is_half", "True"))
+from config import is_half
 import sys, numpy as np, traceback, pdb
 import os.path
 from glob import glob

GPT_SoVITS/prepare_datasets/2-get-hubert-wav32k.py

@@ -10,7 +10,7 @@ os.environ["CUDA_VISIBLE_DEVICES"]= os.environ.get("_CUDA_VISIBLE_DEVICES")
 from feature_extractor import cnhubert
 opt_dir=                            os.environ.get("opt_dir")
 cnhubert.cnhubert_base_path=                os.environ.get("cnhubert_base_dir")
-is_half=eval(os.environ.get("is_half","True"))
+from config import is_half
 
 import pdb,traceback,numpy as np,logging
 from scipy.io import wavfile

GPT_SoVITS/prepare_datasets/3-get-semantic.py

@@ -8,7 +8,7 @@ os.environ["CUDA_VISIBLE_DEVICES"] = os.environ.get("_CUDA_VISIBLE_DEVICES")
 opt_dir = os.environ.get("opt_dir")
 pretrained_s2G = os.environ.get("pretrained_s2G")
 s2config_path = os.environ.get("s2config_path")
-is_half = eval(os.environ.get("is_half", "True"))
+from config import is_half
 import math, traceback
 import multiprocessing
 import sys, pdb

config.py

@@ -42,6 +42,7 @@ if infer_device == "cuda":
         is_half=False
 
 if(infer_device=="cpu"):is_half=False
+if(torch.backends.mps.is_available()):is_half=False
 
 class Config:
     def __init__(self):

tools/cmd-denoise.py

@@ -1,29 +1,29 @@
-import os,argparse
+import os,argparse
-
+
-from modelscope.pipelines import pipeline
+from modelscope.pipelines import pipeline
-from modelscope.utils.constant import Tasks
+from modelscope.utils.constant import Tasks
-from tqdm import tqdm
+from tqdm import tqdm
-
+
-path_denoise  = 'tools/denoise-model/speech_frcrn_ans_cirm_16k'
+path_denoise  = 'tools/denoise-model/speech_frcrn_ans_cirm_16k'
-path_denoise  = path_denoise  if os.path.exists(path_denoise)  else "damo/speech_frcrn_ans_cirm_16k"
+path_denoise  = path_denoise  if os.path.exists(path_denoise)  else "damo/speech_frcrn_ans_cirm_16k"
-ans = pipeline(Tasks.acoustic_noise_suppression,model=path_denoise)
+ans = pipeline(Tasks.acoustic_noise_suppression,model=path_denoise)
-def execute_denoise(input_folder,output_folder):
+def execute_denoise(input_folder,output_folder):
-    os.makedirs(output_folder,exist_ok=True)
+    os.makedirs(output_folder,exist_ok=True)
-    # print(input_folder)
+    # print(input_folder)
-    # print(list(os.listdir(input_folder).sort()))
+    # print(list(os.listdir(input_folder).sort()))
-    for name in tqdm(os.listdir(input_folder)):
+    for name in tqdm(os.listdir(input_folder)):
-        ans("%s/%s"%(input_folder,name),output_path='%s/%s'%(output_folder,name))
+        ans("%s/%s"%(input_folder,name),output_path='%s/%s'%(output_folder,name))
-
+
-if __name__ == '__main__':
+if __name__ == '__main__':
-    parser = argparse.ArgumentParser()
+    parser = argparse.ArgumentParser()
-    parser.add_argument("-i", "--input_folder", type=str, required=True,
+    parser.add_argument("-i", "--input_folder", type=str, required=True,
-                        help="Path to the folder containing WAV files.")
+                        help="Path to the folder containing WAV files.")
-    parser.add_argument("-o", "--output_folder", type=str, required=True, 
+    parser.add_argument("-o", "--output_folder", type=str, required=True, 
-                        help="Output folder to store transcriptions.")
+                        help="Output folder to store transcriptions.")
-    parser.add_argument("-p", "--precision", type=str, default='float16', choices=['float16','float32'],
+    parser.add_argument("-p", "--precision", type=str, default='float16', choices=['float16','float32'],
-                        help="fp16 or fp32")#还没接入
+                        help="fp16 or fp32")#还没接入
-    cmd = parser.parse_args()
+    cmd = parser.parse_args()
-    execute_denoise(
+    execute_denoise(
-        input_folder  = cmd.input_folder,
+        input_folder  = cmd.input_folder,
-        output_folder = cmd.output_folder,
+        output_folder = cmd.output_folder,
     )