Merge branch 'main' of https://github.com/pengooseDev/GPT-SoVITS

tools/asr/config.py

@@ -0,0 +1,31 @@
+import os
+
+def check_fw_local_models():
+    '''
+    启动时检查本地是否有 Faster Whisper 模型.
+    '''
+    model_size_list = [
+        "tiny",     "tiny.en", 
+        "base",     "base.en", 
+        "small",    "small.en", 
+        "medium",   "medium.en", 
+        "large",    "large-v1", 
+        "large-v2", "large-v3"]
+    for i, size in enumerate(model_size_list):
+        if os.path.exists(f'tools/asr/models/faster-whisper-{size}'):
+            model_size_list[i] = size + '-local'
+    return model_size_list
+
+asr_dict = {
+    "达摩 ASR (中文)": {
+        'lang': ['zh'],
+        'size': ['large'],
+        'path': 'funasr_asr.py',
+    },
+    "Faster Whisper (多语种)": {
+        'lang': ['auto', 'zh', 'en', 'ja'],
+        'size': check_fw_local_models(),
+        'path': 'fasterwhisper_asr.py'
+    }
+}
+

tools/asr/fasterwhisper_asr.py

@@ -0,0 +1,107 @@
+import argparse
+import os
+os.environ["HF_ENDPOINT"]="https://hf-mirror.com"
+import traceback
+import requests
+from glob import glob
+
+from faster_whisper import WhisperModel
+from tqdm import tqdm
+
+from tools.asr.config import check_fw_local_models
+from tools.asr.funasr_asr import only_asr
+
+os.environ["KMP_DUPLICATE_LIB_OK"]="TRUE"
+
+language_code_list = [
+    "af", "am", "ar", "as", "az", 
+    "ba", "be", "bg", "bn", "bo", 
+    "br", "bs", "ca", "cs", "cy", 
+    "da", "de", "el", "en", "es", 
+    "et", "eu", "fa", "fi", "fo", 
+    "fr", "gl", "gu", "ha", "haw", 
+    "he", "hi", "hr", "ht", "hu", 
+    "hy", "id", "is", "it", "ja", 
+    "jw", "ka", "kk", "km", "kn", 
+    "ko", "la", "lb", "ln", "lo", 
+    "lt", "lv", "mg", "mi", "mk", 
+    "ml", "mn", "mr", "ms", "mt", 
+    "my", "ne", "nl", "nn", "no", 
+    "oc", "pa", "pl", "ps", "pt", 
+    "ro", "ru", "sa", "sd", "si", 
+    "sk", "sl", "sn", "so", "sq", 
+    "sr", "su", "sv", "sw", "ta", 
+    "te", "tg", "th", "tk", "tl", 
+    "tr", "tt", "uk", "ur", "uz", 
+    "vi", "yi", "yo", "zh", "yue",
+    "auto"]
+
+def execute_asr(input_folder, output_folder, model_size, language,precision):
+    if '-local' in model_size:
+        model_size = model_size[:-6]
+        model_path = f'tools/asr/models/faster-whisper-{model_size}'
+    else:
+        model_path = model_size
+    if language == 'auto':
+        language = None #不设置语种由模型自动输出概率最高的语种
+    print("loading faster whisper model:",model_size,model_path)
+    try:
+        model = WhisperModel(model_path, device="cuda", compute_type=precision)
+    except:
+        return print(traceback.format_exc())
+    output = []
+    output_file_name = os.path.basename(input_folder)
+    output_file_path = os.path.abspath(f'{output_folder}/{output_file_name}.list')
+
+    if not os.path.exists(output_folder):
+        os.makedirs(output_folder)
+
+    for file in tqdm(glob(os.path.join(input_folder, '**/*.wav'), recursive=True)):
+        try:
+            segments, info = model.transcribe(
+                audio          = file,
+                beam_size      = 5,
+                vad_filter     = True,
+                vad_parameters = dict(min_silence_duration_ms=700),
+                language       = language)
+            text = ''
+
+            if info.language == "zh":
+                print("检测为中文文本,转funasr处理")
+                text = only_asr(file)
+
+            if text == '':
+                for segment in segments:
+                    text += segment.text
+            output.append(f"{file}|{output_file_name}|{info.language.upper()}|{text}")
+        except:
+            return print(traceback.format_exc())
+        
+    with open(output_file_path, "w", encoding="utf-8") as f:
+        f.write("\n".join(output))
+        print(f"ASR 任务完成->标注文件路径: {output_file_path}\n")
+    return output_file_path
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser()
+    parser.add_argument("-i", "--input_folder", type=str, required=True,
+                        help="Path to the folder containing WAV files.")
+    parser.add_argument("-o", "--output_folder", type=str, required=True, 
+                        help="Output folder to store transcriptions.")
+    parser.add_argument("-s", "--model_size", type=str, default='large-v3', 
+                        choices=check_fw_local_models(),
+                        help="Model Size of Faster Whisper")
+    parser.add_argument("-l", "--language", type=str, default='ja',
+                        choices=language_code_list,
+                        help="Language of the audio files.")
+    parser.add_argument("-p", "--precision", type=str, default='float16', choices=['float16','float32'],
+                        help="fp16 or fp32")
+
+    cmd = parser.parse_args()
+    output_file_path = execute_asr(
+        input_folder  = cmd.input_folder,
+        output_folder = cmd.output_folder,
+        model_size    = cmd.model_size,
+        language      = cmd.language,
+        precision     = cmd.precision,
+    )

tools/asr/funasr_asr.py

@@ -0,0 +1,76 @@
+# -*- coding:utf-8 -*-
+
+import argparse
+import os
+import traceback
+from tqdm import tqdm
+
+from funasr import AutoModel
+
+path_asr  = 'tools/asr/models/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch'
+path_vad  = 'tools/asr/models/speech_fsmn_vad_zh-cn-16k-common-pytorch'
+path_punc = 'tools/asr/models/punc_ct-transformer_zh-cn-common-vocab272727-pytorch'
+path_asr  = path_asr  if os.path.exists(path_asr)  else "iic/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch"
+path_vad  = path_vad  if os.path.exists(path_vad)  else "iic/speech_fsmn_vad_zh-cn-16k-common-pytorch"
+path_punc = path_punc if os.path.exists(path_punc) else "iic/punc_ct-transformer_zh-cn-common-vocab272727-pytorch"
+
+model = AutoModel(
+    model               = path_asr,
+    model_revision      = "v2.0.4",
+    vad_model           = path_vad,
+    vad_model_revision  = "v2.0.4",
+    punc_model          = path_punc,
+    punc_model_revision = "v2.0.4",
+)
+
+def only_asr(input_file):
+    try:
+        text = model.generate(input=input_file)[0]["text"]
+    except:
+        text = ''
+        print(traceback.format_exc())
+    return text
+
+def execute_asr(input_folder, output_folder, model_size, language):
+    input_file_names = os.listdir(input_folder)
+    input_file_names.sort()
+    
+    output = []
+    output_file_name = os.path.basename(input_folder)
+
+    for name in tqdm(input_file_names):
+        try:
+            text = model.generate(input="%s/%s"%(input_folder, name))[0]["text"]
+            output.append(f"{input_folder}/{name}|{output_file_name}|{language.upper()}|{text}")
+        except:
+            print(traceback.format_exc())
+
+    output_folder = output_folder or "output/asr_opt"
+    os.makedirs(output_folder, exist_ok=True)
+    output_file_path = os.path.abspath(f'{output_folder}/{output_file_name}.list')
+
+    with open(output_file_path, "w", encoding="utf-8") as f:
+        f.write("\n".join(output))
+        print(f"ASR 任务完成->标注文件路径: {output_file_path}\n")
+    return output_file_path
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser()
+    parser.add_argument("-i", "--input_folder", type=str, required=True,
+                        help="Path to the folder containing WAV files.")
+    parser.add_argument("-o", "--output_folder", type=str, required=True, 
+                        help="Output folder to store transcriptions.")
+    parser.add_argument("-s", "--model_size", type=str, default='large',
+                        help="Model Size of FunASR is Large")
+    parser.add_argument("-l", "--language", type=str, default='zh', choices=['zh'],
+                        help="Language of the audio files.")
+    parser.add_argument("-p", "--precision", type=str, default='float16', choices=['float16','float32'],
+                        help="fp16 or fp32")#还没接入
+
+    cmd = parser.parse_args()
+    execute_asr(
+        input_folder  = cmd.input_folder,
+        output_folder = cmd.output_folder,
+        model_size    = cmd.model_size,
+        language      = cmd.language,
+    )

tools/asr/models/.gitignore

@@ -0,0 +1,2 @@
+*
+!.gitignore

tools/damo_asr/cmd-asr.py

@@ -1,27 +0,0 @@
-# -*- coding:utf-8 -*-
-
-from modelscope.pipelines import pipeline
-from modelscope.utils.constant import Tasks
-import sys,os,traceback
-dir=sys.argv[1]
-# opt_name=dir.split("\\")[-1].split("/")[-1]
-opt_name=os.path.basename(dir)
-inference_pipeline = pipeline(
-    task=Tasks.auto_speech_recognition,
-    model='tools/damo_asr/models/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch',
-    vad_model='tools/damo_asr/models/speech_fsmn_vad_zh-cn-16k-common-pytorch',
-    punc_model='tools/damo_asr/models/punc_ct-transformer_zh-cn-common-vocab272727-pytorch',
-)
-
-opt=[]
-for name in os.listdir(dir):
-    try:
-        text = inference_pipeline(audio_in="%s/%s"%(dir,name))["text"]
-        opt.append("%s/%s|%s|ZH|%s"%(dir,name,opt_name,text))
-    except:
-        print(traceback.format_exc())
-
-opt_dir="output/asr_opt"
-os.makedirs(opt_dir,exist_ok=True)
-with open("%s/%s.list"%(opt_dir,opt_name),"w",encoding="utf-8")as f:f.write("\n".join(opt))
-

tools/my_utils.py

@@ -1,3 +1,4 @@
+import platform,os,traceback
 import ffmpeg
 import numpy as np
 
@@ -7,15 +8,24 @@ def load_audio(file, sr):
         # https://github.com/openai/whisper/blob/main/whisper/audio.py#L26
         # This launches a subprocess to decode audio while down-mixing and resampling as necessary.
         # Requires the ffmpeg CLI and `ffmpeg-python` package to be installed.
-        file = (
-            file.strip(" ").strip('"').strip("\n").strip('"').strip(" ")
-        )  # 防止小白拷路径头尾带了空格和"和回车
+        file = clean_path(file)  # 防止小白拷路径头尾带了空格和"和回车
+        if os.path.exists(file) == False:
+            raise RuntimeError(
+                "You input a wrong audio path that does not exists, please fix it!"
+            )
         out, _ = (
             ffmpeg.input(file, threads=0)
             .output("-", format="f32le", acodec="pcm_f32le", ac=1, ar=sr)
             .run(cmd=["ffmpeg", "-nostdin"], capture_stdout=True, capture_stderr=True)
         )
     except Exception as e:
+        traceback.print_exc()
         raise RuntimeError(f"Failed to load audio: {e}")
 
     return np.frombuffer(out, np.float32).flatten()
+
+
+def clean_path(path_str):
+    if platform.system() == 'Windows':
+        path_str = path_str.replace('/', '\\')
+    return path_str.strip(" ").strip('"').strip("\n").strip('"').strip(" ")

tools/slice_audio.py

@@ -11,7 +11,7 @@ def slice(inp,opt_root,threshold,min_length,min_interval,hop_size,max_sil_kept,_
     if os.path.isfile(inp):
         input=[inp]
     elif os.path.isdir(inp):
-        input=["%s/%s"%(inp,name)for name in sorted(list(os.listdir(inp)))]
+        input=[os.path.join(inp, name) for name in sorted(list(os.listdir(inp)))]
     else:
         return "输入路径存在但既不是文件也不是文件夹"
     slicer = Slicer(
@@ -35,7 +35,7 @@ def slice(inp,opt_root,threshold,min_length,min_interval,hop_size,max_sil_kept,_
                 if(tmp_max>1):chunk/=tmp_max
                 chunk = (chunk / tmp_max * (_max * alpha)) + (1 - alpha) * chunk
                 wavfile.write(
-                    "%s/%s_%s_%s.wav" % (opt_root, name, start, end),
+                    "%s/%s_%010d_%010d.wav" % (opt_root, name, start, end),
                     32000,
                     # chunk.astype(np.float32),
                     (chunk * 32767).astype(np.int16),

tools/subfix_webui.py

@@ -79,6 +79,7 @@ def b_change_index(index, batch):
 
 
 def b_next_index(index, batch):
+    b_save_file()
     if (index + batch) <= g_max_json_index:
         return index + batch , *b_change_index(index + batch, batch)
     else:
@@ -86,6 +87,7 @@ def b_next_index(index, batch):
 
 
 def b_previous_index(index, batch):
+    b_save_file()
     if (index - batch) >= 0:
         return index - batch , *b_change_index(index - batch, batch)
     else:
@@ -108,6 +110,7 @@ def b_submit_change(*text_list):
 
 def b_delete_audio(*checkbox_list):
     global g_data_json, g_index, g_max_json_index
+    b_save_file()
     change = False
     for i, checkbox in reversed(list(enumerate(checkbox_list))):
         if g_index + i < len(g_data_json):
@@ -119,8 +122,8 @@ def b_delete_audio(*checkbox_list):
     if g_index > g_max_json_index:
         g_index = g_max_json_index
         g_index = g_index if g_index >= 0 else 0
-    # if change:
-    #     b_save_file()
+    if change:
+        b_save_file()
     # return gr.Slider(value=g_index, maximum=(g_max_json_index if g_max_json_index>=0 else 0)), *b_change_index(g_index, g_batch)
     return {"value":g_index,"__type__":"update","maximum":(g_max_json_index if g_max_json_index>=0 else 0)},*b_change_index(g_index, g_batch)
 
@@ -170,6 +173,7 @@ def b_audio_split(audio_breakpoint, *checkbox_list):
 
 def b_merge_audio(interval_r, *checkbox_list):
     global g_data_json , g_max_json_index
+    b_save_file()
     checked_index = []
     audios_path = []
     audios_text = []
@@ -294,6 +298,7 @@ def set_global(load_json, load_list, json_key_text, json_key_path, batch):
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(description='Process some integers.')
     parser.add_argument('--load_json', default="None", help='source file, like demo.json')
+    parser.add_argument('--is_share', default="False", help='whether webui is_share=True')
     parser.add_argument('--load_list', default="None", help='source file, like demo.list')
     parser.add_argument('--webui_port_subfix', default=9871, help='source file, like demo.list')
     parser.add_argument('--json_key_text', default="text", help='the text key name in json, Default: text')
@@ -488,5 +493,6 @@ if __name__ == "__main__":
         server_name="0.0.0.0",
         inbrowser=True,
         quiet=True,
+        share=eval(args.is_share),
         server_port=int(args.webui_port_subfix)
     )

tools/uvr5/lib/lib_v5/spec_utils.py

@@ -43,8 +43,8 @@ def wave_to_spectrogram(
         wave_left = np.asfortranarray(wave[0])
         wave_right = np.asfortranarray(wave[1])
 
-    spec_left = librosa.stft(wave_left, n_fft, hop_length=hop_length)
-    spec_right = librosa.stft(wave_right, n_fft, hop_length=hop_length)
+    spec_left  = librosa.stft(wave_left,  n_fft=n_fft, hop_length=hop_length)
+    spec_right = librosa.stft(wave_right, n_fft=n_fft, hop_length=hop_length)
 
     spec = np.asfortranarray([spec_left, spec_right])
 
@@ -78,7 +78,7 @@ def wave_to_spectrogram_mt(
         kwargs={"y": wave_left, "n_fft": n_fft, "hop_length": hop_length},
     )
     thread.start()
-    spec_right = librosa.stft(wave_right, n_fft, hop_length=hop_length)
+    spec_right = librosa.stft(wave_right, n_fft=n_fft, hop_length=hop_length)
     thread.join()
 
     spec = np.asfortranarray([spec_left, spec_right])
@@ -230,27 +230,31 @@ def cache_or_load(mix_path, inst_path, mp):
 
             if d == len(mp.param["band"]):  # high-end band
                 X_wave[d], _ = librosa.load(
-                    mix_path, bp["sr"], False, dtype=np.float32, res_type=bp["res_type"]
+                    mix_path, 
+                    sr       = bp["sr"],
+                    mono     = False,
+                    dtype    = np.float32,
+                    res_type = bp["res_type"]
                 )
                 y_wave[d], _ = librosa.load(
                     inst_path,
-                    bp["sr"],
-                    False,
-                    dtype=np.float32,
-                    res_type=bp["res_type"],
+                    sr       = bp["sr"],
+                    mono     = False,
+                    dtype    = np.float32,
+                    res_type = bp["res_type"],
                 )
             else:  # lower bands
                 X_wave[d] = librosa.resample(
                     X_wave[d + 1],
-                    mp.param["band"][d + 1]["sr"],
-                    bp["sr"],
-                    res_type=bp["res_type"],
+                    orig_sr   = mp.param["band"][d + 1]["sr"],
+                    target_sr = bp["sr"],
+                    res_type  = bp["res_type"],
                 )
                 y_wave[d] = librosa.resample(
                     y_wave[d + 1],
-                    mp.param["band"][d + 1]["sr"],
-                    bp["sr"],
-                    res_type=bp["res_type"],
+                    orig_sr   = mp.param["band"][d + 1]["sr"],
+                    target_sr = bp["sr"],
+                    res_type  = bp["res_type"],
                 )
 
             X_wave[d], y_wave[d] = align_wave_head_and_tail(X_wave[d], y_wave[d])
@@ -401,9 +405,9 @@ def cmb_spectrogram_to_wave(spec_m, mp, extra_bins_h=None, extra_bins=None):
                         mp.param["mid_side_b2"],
                         mp.param["reverse"],
                     ),
-                    bp["sr"],
-                    sr,
-                    res_type="sinc_fastest",
+                    orig_sr   = bp["sr"],
+                    target_sr = sr,
+                    res_type  = "sinc_fastest",
                 )
             else:  # mid
                 spec_s = fft_hp_filter(spec_s, bp["hpf_start"], bp["hpf_stop"] - 1)
@@ -418,8 +422,8 @@ def cmb_spectrogram_to_wave(spec_m, mp, extra_bins_h=None, extra_bins=None):
                         mp.param["reverse"],
                     ),
                 )
-                # wave = librosa.core.resample(wave2, bp['sr'], sr, res_type="sinc_fastest")
-                wave = librosa.core.resample(wave2, bp["sr"], sr, res_type="scipy")
+                # wave = librosa.core.resample(wave2, orig_sr=bp['sr'], target_sr=sr, res_type="sinc_fastest")
+                wave = librosa.core.resample(wave2, orig_sr=bp["sr"], target_sr=sr, res_type="scipy")
 
     return wave.T
 
@@ -506,8 +510,8 @@ def ensembling(a, specs):
 def stft(wave, nfft, hl):
     wave_left = np.asfortranarray(wave[0])
     wave_right = np.asfortranarray(wave[1])
-    spec_left = librosa.stft(wave_left, nfft, hop_length=hl)
-    spec_right = librosa.stft(wave_right, nfft, hop_length=hl)
+    spec_left = librosa.stft(wave_left, n_fft=nfft, hop_length=hl)
+    spec_right = librosa.stft(wave_right, n_fft=nfft, hop_length=hl)
     spec = np.asfortranarray([spec_left, spec_right])
 
     return spec
@@ -569,10 +573,10 @@ if __name__ == "__main__":
             if d == len(mp.param["band"]):  # high-end band
                 wave[d], _ = librosa.load(
                     args.input[i],
-                    bp["sr"],
-                    False,
-                    dtype=np.float32,
-                    res_type=bp["res_type"],
+                    sr       = bp["sr"],
+                    mono     = False,
+                    dtype    = np.float32,
+                    res_type = bp["res_type"],
                 )
 
                 if len(wave[d].shape) == 1:  # mono to stereo
@@ -580,9 +584,9 @@ if __name__ == "__main__":
             else:  # lower bands
                 wave[d] = librosa.resample(
                     wave[d + 1],
-                    mp.param["band"][d + 1]["sr"],
-                    bp["sr"],
-                    res_type=bp["res_type"],
+                    orig_sr   = mp.param["band"][d + 1]["sr"],
+                    target_sr = bp["sr"],
+                    res_type  = bp["res_type"],
                 )
 
             spec[d] = wave_to_spectrogram(

tools/uvr5/lib/utils.py

@@ -24,7 +24,7 @@ def make_padding(width, cropsize, offset):
 
 def inference(X_spec, device, model, aggressiveness, data):
     """
-    data ： dic configs
+    data : dic configs
     """
 
     def _execute(

tools/uvr5/mdxnet.py

@@ -239,7 +239,7 @@ class Predictor:
 
 
 class MDXNetDereverb:
-    def __init__(self, chunks, device):
+    def __init__(self, chunks):
         self.onnx = "%s/uvr5_weights/onnx_dereverb_By_FoxJoy"%os.path.dirname(os.path.abspath(__file__))
         self.shifts = 10  # 'Predict with randomised equivariant stabilisation'
         self.mixing = "min_mag"  # ['default','min_mag','max_mag']
@@ -250,7 +250,7 @@ class MDXNetDereverb:
         self.n_fft = 6144
         self.denoise = True
         self.pred = Predictor(self)
-        self.device = device
+        self.device = cpu
 
     def _path_audio_(self, input, vocal_root, others_root, format, is_hp3=False):
         self.pred.prediction(input, vocal_root, others_root, format)

tools/uvr5/vr.py

@@ -61,19 +61,19 @@ class AudioPre:
                     _,
                 ) = librosa.core.load(  # 理论上librosa读取可能对某些音频有bug，应该上ffmpeg读取，但是太麻烦了弃坑
                     music_file,
-                    bp["sr"],
-                    False,
-                    dtype=np.float32,
-                    res_type=bp["res_type"],
+                    sr       = bp["sr"],
+                    mono     = False,
+                    dtype    = np.float32,
+                    res_type = bp["res_type"],
                 )
                 if X_wave[d].ndim == 1:
                     X_wave[d] = np.asfortranarray([X_wave[d], X_wave[d]])
             else:  # lower bands
                 X_wave[d] = librosa.core.resample(
                     X_wave[d + 1],
-                    self.mp.param["band"][d + 1]["sr"],
-                    bp["sr"],
-                    res_type=bp["res_type"],
+                    orig_sr   = self.mp.param["band"][d + 1]["sr"],
+                    target_sr = bp["sr"],
+                    res_type  = bp["res_type"],
                 )
             # Stft of wave source
             X_spec_s[d] = spec_utils.wave_to_spectrogram_mt(
@@ -110,6 +110,9 @@ class AudioPre:
         y_spec_m = pred * X_phase
         v_spec_m = X_spec_m - y_spec_m
 
+        if is_hp3 == True:
+            ins_root,vocal_root = vocal_root,ins_root
+
         if ins_root is not None:
             if self.data["high_end_process"].startswith("mirroring"):
                 input_high_end_ = spec_utils.mirroring(
@@ -242,19 +245,19 @@ class AudioPreDeEcho:
                     _,
                 ) = librosa.core.load(  # 理论上librosa读取可能对某些音频有bug，应该上ffmpeg读取，但是太麻烦了弃坑
                     music_file,
-                    bp["sr"],
-                    False,
-                    dtype=np.float32,
-                    res_type=bp["res_type"],
+                    sr       = bp["sr"],
+                    mono     = False,
+                    dtype    = np.float32,
+                    res_type = bp["res_type"],
                 )
                 if X_wave[d].ndim == 1:
                     X_wave[d] = np.asfortranarray([X_wave[d], X_wave[d]])
             else:  # lower bands
                 X_wave[d] = librosa.core.resample(
                     X_wave[d + 1],
-                    self.mp.param["band"][d + 1]["sr"],
-                    bp["sr"],
-                    res_type=bp["res_type"],
+                    orig_sr   = self.mp.param["band"][d + 1]["sr"],
+                    target_sr = bp["sr"],
+                    res_type  = bp["res_type"],
                 )
             # Stft of wave source
             X_spec_s[d] = spec_utils.wave_to_spectrogram_mt(

tools/uvr5/webui.py

@@ -5,7 +5,8 @@ from tools.i18n.i18n import I18nAuto
 i18n = I18nAuto()
 
 logger = logging.getLogger(__name__)
-import ffmpeg
+import librosa,ffmpeg
+import soundfile as sf
 import torch
 import sys
 from mdxnet import MDXNetDereverb
@@ -18,8 +19,9 @@ for name in os.listdir(weight_uvr5_root):
         uvr5_names.append(name.replace(".pth", ""))
 
 device=sys.argv[1]
-is_half=sys.argv[2]
-
+is_half=eval(sys.argv[2])
+webui_port_uvr5=int(sys.argv[3])
+is_share=eval(sys.argv[4])
 
 def uvr(model_name, inp_root, save_root_vocal, paths, save_root_ins, agg, format0):
     infos = []
@@ -31,25 +33,24 @@ def uvr(model_name, inp_root, save_root_vocal, paths, save_root_ins, agg, format
         save_root_ins = (
             save_root_ins.strip(" ").strip('"').strip("\n").strip('"').strip(" ")
         )
+        is_hp3 = "HP3" in model_name
         if model_name == "onnx_dereverb_By_FoxJoy":
-            pre_fun = MDXNetDereverb(15, device)
+            pre_fun = MDXNetDereverb(15)
         else:
             func = AudioPre if "DeEcho" not in model_name else AudioPreDeEcho
             pre_fun = func(
                 agg=int(agg),
-                model_path=os.path.join(
-                    weight_uvr5_root, model_name + ".pth"
-                ),
+                model_path=os.path.join(weight_uvr5_root, model_name + ".pth"),
                 device=device,
                 is_half=is_half,
             )
-        is_hp3 = "HP3" in model_name
         if inp_root != "":
             paths = [os.path.join(inp_root, name) for name in os.listdir(inp_root)]
         else:
             paths = [path.name for path in paths]
         for path in paths:
             inp_path = os.path.join(inp_root, path)
+            if(os.path.isfile(inp_path)==False):continue
             need_reformat = 1
             done = 0
             try:
@@ -60,7 +61,7 @@ def uvr(model_name, inp_root, save_root_vocal, paths, save_root_ins, agg, format
                 ):
                     need_reformat = 0
                     pre_fun._path_audio_(
-                        inp_path, save_root_ins, save_root_vocal, format0, is_hp3=is_hp3
+                        inp_path, save_root_ins, save_root_vocal, format0,is_hp3
                     )
                     done = 1
             except:
@@ -79,23 +80,15 @@ def uvr(model_name, inp_root, save_root_vocal, paths, save_root_ins, agg, format
             try:
                 if done == 0:
                     pre_fun._path_audio_(
-                        inp_path, save_root_ins, save_root_vocal, format0
+                        inp_path, save_root_ins, save_root_vocal, format0,is_hp3
                     )
                 infos.append("%s->Success" % (os.path.basename(inp_path)))
                 yield "\n".join(infos)
             except:
-                try:
-                    if done == 0:
-                        pre_fun._path_audio_(
-                            inp_path, save_root_ins, save_root_vocal, format0
-                        )
-                    infos.append("%s->Success" % (os.path.basename(inp_path)))
-                    yield "\n".join(infos)
-                except:
-                    infos.append(
-                        "%s->%s" % (os.path.basename(inp_path), traceback.format_exc())
-                    )
-                    yield "\n".join(infos)
+                infos.append(
+                    "%s->%s" % (os.path.basename(inp_path), traceback.format_exc())
+                )
+                yield "\n".join(infos)
     except:
         infos.append(traceback.format_exc())
         yield "\n".join(infos)
@@ -109,16 +102,15 @@ def uvr(model_name, inp_root, save_root_vocal, paths, save_root_ins, agg, format
                 del pre_fun
         except:
             traceback.print_exc()
+        print("clean_empty_cache")
         if torch.cuda.is_available():
             torch.cuda.empty_cache()
-            logger.info("Executed torch.cuda.empty_cache()")
     yield "\n".join(infos)
 
-
-with gr.Blocks(title="RVC WebUI") as app:
+with gr.Blocks(title="UVR5 WebUI") as app:
     gr.Markdown(
         value=
-            "本软件以MIT协议开源, 作者不对软件具备任何控制力, 使用软件者、传播软件导出的声音者自负全责. <br>如不认可该条款, 则不能使用或引用软件包内任何代码和文件. 详见根目录<b>LICENSE</b>."
+            i18n("本软件以MIT协议开源, 作者不对软件具备任何控制力, 使用软件者、传播软件导出的声音者自负全责. <br>如不认可该条款, 则不能使用或引用软件包内任何代码和文件. 详见根目录<b>LICENSE</b>.")
     )
     with gr.Tabs():
         with gr.TabItem(i18n("伴奏人声分离&去混响&去回声")):
@@ -143,7 +135,7 @@ with gr.Blocks(title="RVC WebUI") as app:
                             minimum=0,
                             maximum=20,
                             step=1,
-                            label="人声提取激进程度",
+                            label=i18n("人声提取激进程度"),
                             value=10,
                             interactive=True,
                             visible=False,  # 先不开放调整
@@ -179,6 +171,7 @@ with gr.Blocks(title="RVC WebUI") as app:
 app.queue(concurrency_count=511, max_size=1022).launch(
     server_name="0.0.0.0",
     inbrowser=True,
-    server_port=9873,
+    share=is_share,
+    server_port=webui_port_uvr5,
     quiet=True,
-)
+)