roll back

more features

README.md

@@ -147,7 +147,17 @@ Users in China region can download these two models by entering the links below
 
 - [UVR5 Weights](https://www.icloud.com.cn/iclouddrive/0bekRKDiJXboFhbfm3lM2fVbA#UVR5_Weights)
 
-For Multilingual ASR, download models from [FunAudioLLM/SenseVoiceSmall](https://huggingface.co/FunAudioLLM/SenseVoiceSmall/tree/main) or [iic/SenseVoiceSmall](https://modelscope.cn/models/iic/SenseVoiceSmall/files) and place them in `tools/asr/models`.
+For Chinese ASR (additionally), download models from [Damo ASR Model](https://modelscope.cn/models/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/files), [Damo VAD Model](https://modelscope.cn/models/damo/speech_fsmn_vad_zh-cn-16k-common-pytorch/files), and [Damo Punc Model](https://modelscope.cn/models/damo/punc_ct-transformer_zh-cn-common-vocab272727-pytorch/files) and place them in `tools/asr/models`.
+
+For English or Japanese ASR (additionally), download models from [Faster Whisper Large V3](https://huggingface.co/Systran/faster-whisper-large-v3) and place them in `tools/asr/models`. Also, [other models](https://huggingface.co/Systran) may have the similar effect with smaller disk footprint. 
+
+Users in China region can download this model by entering the links below
+
+- [Faster Whisper Large V3](https://www.icloud.com/iclouddrive/0c4pQxFs7oWyVU1iMTq2DbmLA#faster-whisper-large-v3) (clicking "Download a copy")
+
+- [Faster Whisper Large V3](https://hf-mirror.com/Systran/faster-whisper-large-v3) (HuggingFace mirror site)
+
+For SenseVoice Multilingual ASR, download models from [FunAudioLLM/SenseVoiceSmall](https://huggingface.co/FunAudioLLM/SenseVoiceSmall/tree/main) or [iic/SenseVoiceSmall](https://modelscope.cn/models/iic/SenseVoiceSmall/files) and place them in `tools/asr/models`.
 
 ## Dataset Format
 
@@ -208,13 +218,23 @@ python audio_slicer.py \
     --min_interval <shortest_time_gap_between_adjacent_subclips> 
     --hop_size <step_size_for_computing_volume_curve>
 ```
-This is how dataset ASR processing is done using the command line
+This is how dataset ASR processing is done using the command line(Only Chinese)
+```
+python tools/asr/funasr_asr.py -i <input> -o <output>
+```
+ASR processing is performed through Faster_Whisper(ASR marking except Chinese)
+
+(No progress bars, GPU performance may cause time delays)
+```
+python ./tools/asr/fasterwhisper_asr.py -i <input> -o <output> -l <language> -p <precision>
+```
+SenseVoice Multilingual ASR
 ```
 python tools/asr/sensevoice.py -i <input> -o <output> -l <language> -d <device>
 ```
-
 A custom list save path is enabled
 
+
 ## Credits
 
 Special thanks to the following projects and contributors:
@@ -239,6 +259,8 @@ Special thanks to the following projects and contributors:
 - [SubFix](https://github.com/cronrpc/SubFix)
 - [FFmpeg](https://github.com/FFmpeg/FFmpeg)
 - [gradio](https://github.com/gradio-app/gradio)
+- [faster-whisper](https://github.com/SYSTRAN/faster-whisper)
+- [FunASR](https://github.com/alibaba-damo-academy/FunASR)
 - [SenseVoice](https://github.com/FunAudioLLM/SenseVoice)
   
 ## Thanks to all contributors for their efforts

docs/cn/README.md

@@ -147,7 +147,17 @@ docker run --rm -it --gpus=all --env=is_half=False --volume=G:\GPT-SoVITS-Docker
 
 - [UVR5 Weights](https://www.icloud.com.cn/iclouddrive/0bekRKDiJXboFhbfm3lM2fVbA#UVR5_Weights)
 
-对于多语言自动语音识别（附加），从 [FunAudioLLM/SenseVoiceSmall](https://huggingface.co/FunAudioLLM/SenseVoiceSmall/tree/main) 或 [iic/SenseVoiceSmall](https://modelscope.cn/models/iic/SenseVoiceSmall/files) 下载模型，并将它们放置在 `tools/asr/models` 中。
+对于中文自动语音识别（附加），从 [Damo ASR Model](https://modelscope.cn/models/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/files), [Damo VAD Model](https://modelscope.cn/models/damo/speech_fsmn_vad_zh-cn-16k-common-pytorch/files), 和 [Damo Punc Model](https://modelscope.cn/models/damo/punc_ct-transformer_zh-cn-common-vocab272727-pytorch/files) 下载模型，并将它们放置在 `tools/asr/models` 中。
+
+对于SenseVoice多语言自动语音识别（附加），从 [FunAudioLLM/SenseVoiceSmall](https://huggingface.co/FunAudioLLM/SenseVoiceSmall/tree/main) 或 [iic/SenseVoiceSmall](https://modelscope.cn/models/iic/SenseVoiceSmall/files) 下载模型，并将它们放置在 `tools/asr/models` 中。
+
+对于英语与日语自动语音识别（附加）,从 [Faster Whisper Large V3](https://huggingface.co/Systran/faster-whisper-large-v3) 下载模型，并将它们放置在 `tools/asr/models` 中。 此外，[其他模型](https://huggingface.co/Systran)可能具有类似效果，但占用更小的磁盘空间。
+
+中国地区用户可以通过以下链接下载：
+
+- [Faster Whisper Large V3](https://www.icloud.com/iclouddrive/0c4pQxFs7oWyVU1iMTq2DbmLA#faster-whisper-large-v3)(点击“下载副本”)
+  
+- [Faster Whisper Large V3](https://hf-mirror.com/Systran/faster-whisper-large-v3)(Hugging Face镜像站)
 
 
 
@@ -210,7 +220,17 @@ python audio_slicer.py \
     --min_interval <shortest_time_gap_between_adjacent_subclips> 
     --hop_size <step_size_for_computing_volume_curve>
 ````
-这是使用命令行完成数据集ASR处理的方式
+这是使用命令行完成数据集ASR处理的方式（仅限中文）
+````
+python tools/asr/funasr_asr.py -i <input> -o <output>
+````
+通过Faster_Whisper进行ASR处理（除中文之外的ASR标记）
+
+（没有进度条，GPU性能可能会导致时间延迟）
+````
+python ./tools/asr/fasterwhisper_asr.py -i <input> -o <output> -l <language> -p <precision>
+````
+使用SenseVoice进行多语言ASR
 ````
 python tools/asr/sensevoice.py -i <input> -o <output> -l <language> -d <device>
 ````
@@ -241,6 +261,8 @@ python tools/asr/sensevoice.py -i <input> -o <output> -l <language> -d <device>
 - [SubFix](https://github.com/cronrpc/SubFix)
 - [FFmpeg](https://github.com/FFmpeg/FFmpeg)
 - [gradio](https://github.com/gradio-app/gradio)
+- [faster-whisper](https://github.com/SYSTRAN/faster-whisper)
+- [FunASR](https://github.com/alibaba-damo-academy/FunASR)
 - [SenseVoice](https://github.com/FunAudioLLM/SenseVoice)
 
 ## 感谢所有贡献者的努力

docs/ja/README.md

@@ -206,7 +206,7 @@ ASR処理はFaster_Whisperを通じて実行されます(中国語を除くASR
 
 (進行状況バーは表示されません。GPU のパフォーマンスにより時間遅延が発生する可能性があります)
 ```
-python ./tools/asr/fasterwhisper_asr.py -i <input> -o <output> -l <language>
+python ./tools/asr/fasterwhisper_asr.py -i <input> -o <output> -l <language> -p <precision>
 ```
 カスタムリストの保存パスが有効になっています
 
@@ -236,6 +236,7 @@ python ./tools/asr/fasterwhisper_asr.py -i <input> -o <output> -l <language>
 - [gradio](https://github.com/gradio-app/gradio)
 - [faster-whisper](https://github.com/SYSTRAN/faster-whisper)
 - [FunASR](https://github.com/alibaba-damo-academy/FunASR)
+- [SenseVoice](https://github.com/FunAudioLLM/SenseVoice)
 
 ## すべてのコントリビューターに感謝します
 

docs/ko/README.md

@@ -210,7 +210,7 @@ ASR 처리는 Faster_Whisper(중국어를 제외한 ASR 마킹)를 통해 수행
 
 (진행률 표시줄 없음, GPU 성능으로 인해 시간 지연이 발생할 수 있음)
 ```
-python ./tools/asr/fasterwhisper_asr.py -i <input> -o <output> -l <language>
+python ./tools/asr/fasterwhisper_asr.py -i <input> -o <output> -l <language> -p <precision>
 ```
 사용자 정의 목록 저장 경로가 활성화되었습니다.
 
@@ -240,7 +240,7 @@ python ./tools/asr/fasterwhisper_asr.py -i <input> -o <output> -l <language>
 - [gradio](https://github.com/gradio-app/gradio)
 - [faster-whisper](https://github.com/SYSTRAN/faster-whisper)
 - [FunASR](https://github.com/alibaba-damo-academy/FunASR)
-
+- [SenseVoice](https://github.com/FunAudioLLM/SenseVoice)
 
 ## 모든 기여자들에게 감사드립니다 ;)
 

docs/tr/README.md

@@ -250,6 +250,7 @@ python ./tools/asr/fasterwhisper_asr.py -i <girdi> -o <çıktı> -l <dil>
 - [gradio](https://github.com/gradio-app/gradio)
 - [faster-whisper](https://github.com/SYSTRAN/faster-whisper)
 - [FunASR](https://github.com/alibaba-damo-academy/FunASR)
+- [SenseVoice](https://github.com/FunAudioLLM/SenseVoice)
   
 ## Tüm katkıda bulunanlara çabaları için teşekkürler
 

requirements.txt

@@ -24,4 +24,5 @@ psutil
 jieba_fast
 jieba
 LangSegment>=0.2.0
+faster_whisper
 wordsegment

tools/asr/config.py

@@ -0,0 +1,39 @@
+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',
+        'precision': 'float32'
+    },
+    "Faster Whisper (多语种)": {
+        'lang': ['auto', 'zh', 'en', 'ja'],
+        'size': check_fw_local_models(),
+        'path': 'fasterwhisper_asr.py',
+        'precision': ['float32', 'float16', 'int8']
+    },
+    "Sense Voice": {
+        'lang': ['auto', 'zh', 'en', 'ja'],
+        'size': ['small'],
+        'path': 'sensevoice_asr.py',
+        'precision': 'float32'
+    }
+}
+

tools/asr/fasterwhisper_asr.py

@@ -0,0 +1,114 @@
+import argparse
+import os
+import traceback
+
+os.environ["HF_ENDPOINT"]          = "https://hf-mirror.com"
+os.environ["KMP_DUPLICATE_LIB_OK"] = "TRUE"
+
+import torch
+from faster_whisper import WhisperModel
+from tqdm import tqdm
+
+from tools.asr.config import check_fw_local_models
+
+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)
+    device = 'cuda' if torch.cuda.is_available() else 'cpu'
+    try:
+        model = WhisperModel(model_path, device=device, compute_type=precision)
+    except:
+        return print(traceback.format_exc())
+    
+    input_file_names = os.listdir(input_folder)
+    input_file_names.sort()
+
+    output = []
+    output_file_name = os.path.basename(input_folder)
+    
+    for file_name in tqdm(input_file_names):
+        try:
+            file_path = os.path.join(input_folder, file_name)
+            segments, info = model.transcribe(
+                audio          = file_path,
+                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")
+                if("only_asr"not in globals()):
+                    from tools.asr.funasr_asr import \
+                        only_asr  # #如果用英文就不需要导入下载模型
+                text = only_asr(file_path)
+
+            if text == '':
+                for segment in segments:
+                    text += segment.text
+            output.append(f"{file_path}|{output_file_name}|{info.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-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','int8'],
+                        help="fp16, int8 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 +1,77 @@
+# -*- 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 file_name in tqdm(input_file_names):
+        try:
+            file_path = os.path.join(input_folder, file_name)
+            text = model.generate(input=file_path)[0]["text"]
+            output.append(f"{file_path}|{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/sensevoice_asr.py

@@ -63,6 +63,11 @@ if __name__ == '__main__':
                         help="Language of the audio files.")
     parser.add_argument("-d", "--device", type=str, default=None, choices=['cpu','cuda'],
                         help="CPU or CUDA")
+    parser.add_argument("-p", "--precision", type=str, default='float32', choices=['float32'],
+                        help="fp16 or fp32")
+    parser.add_argument("-s", "--model_size", type=str, default='small', 
+                        choices=['small'],
+                        help="Model Size of Faster Whisper")
 
     cmd = parser.parse_args()
     output_file_path = execute_asr(

webui.py

@@ -194,14 +194,17 @@ def change_tts_inference(if_tts,bert_path,cnhubert_base_path,gpu_number,gpt_path
         p_tts_inference=None
         yield i18n("TTS推理进程已关闭")
 
-def open_asr(asr_inp_dir, asr_opt_dir, asr_model, asr_model_size, asr_lang):
+from tools.asr.config import asr_dict
+def open_asr(asr_inp_dir, asr_opt_dir, asr_model, asr_model_size, asr_lang, asr_precision):
     global p_asr
     if(p_asr==None):
         asr_inp_dir=my_utils.clean_path(asr_inp_dir)
-        cmd = f'"{python_exec}" tools/asr/sensevoice.py'
+        cmd = f'"{python_exec}" tools/asr/{asr_dict[asr_model]["path"]}'
         cmd += f' -i "{asr_inp_dir}"'
         cmd += f' -o "{asr_opt_dir}"'
+        cmd += f' -s {asr_model_size}'
         cmd += f' -l {asr_lang}'
+        cmd += f" -p {asr_precision}"
         output_file_name = os.path.basename(asr_inp_dir)
         output_folder = asr_opt_dir or "output/asr_opt"
         output_file_path = os.path.abspath(f'{output_folder}/{output_file_name}.list')
@@ -228,14 +231,14 @@ def open_denoise(denoise_inp_dir, denoise_opt_dir):
         denoise_opt_dir=my_utils.clean_path(denoise_opt_dir)
         cmd = '"%s" tools/cmd-denoise.py -i "%s" -o "%s" -p %s'%(python_exec,denoise_inp_dir,denoise_opt_dir,"float16"if is_half==True else "float32")
 
-        yield "语音降噪任务开启：%s"%cmd,{"__type__":"update","visible":False},{"__type__":"update","visible":True}
+        yield "语音降噪任务开启：%s"%cmd, {"__type__":"update","visible":False}, {"__type__":"update","visible":True}, {"__type__":"update"}
         print(cmd)
         p_denoise = Popen(cmd, shell=True)
         p_denoise.wait()
         p_denoise=None
-        yield f"语音降噪任务完成, 查看终端进行下一步",{"__type__":"update","visible":True},{"__type__":"update","visible":False}
+        yield f"语音降噪任务完成, 查看终端进行下一步", {"__type__":"update","visible":True}, {"__type__":"update","visible":False}, {"__type__":"update","value":denoise_opt_dir}
     else:
-        yield "已有正在进行的语音降噪任务，需先终止才能开启下一次任务",{"__type__":"update","visible":False},{"__type__":"update","visible":True}
+        yield "已有正在进行的语音降噪任务，需先终止才能开启下一次任务", {"__type__":"update","visible":False}, {"__type__":"update","visible":True}, {"__type__":"update"}
         # return None
 
 def close_denoise():
@@ -342,12 +345,12 @@ def open_slice(inp,opt_root,threshold,min_length,min_interval,hop_size,max_sil_k
     inp = my_utils.clean_path(inp)
     opt_root = my_utils.clean_path(opt_root)
     if(os.path.exists(inp)==False):
-        yield "输入路径不存在",{"__type__":"update","visible":True},{"__type__":"update","visible":False}
+        yield "输入路径不存在", {"__type__":"update","visible":True}, {"__type__":"update","visible":False}, {"__type__": "update"}, {"__type__": "update"}
         return
     if os.path.isfile(inp):n_parts=1
     elif os.path.isdir(inp):pass
     else:
-        yield "输入路径存在但既不是文件也不是文件夹",{"__type__":"update","visible":True},{"__type__":"update","visible":False}
+        yield "输入路径存在但既不是文件也不是文件夹", {"__type__":"update","visible":True}, {"__type__":"update","visible":False}, {"__type__": "update"}, {"__type__": "update"}
         return
     if (ps_slice == []):
         for i_part in range(n_parts):
@@ -355,13 +358,13 @@ def open_slice(inp,opt_root,threshold,min_length,min_interval,hop_size,max_sil_k
             print(cmd)
             p = Popen(cmd, shell=True)
             ps_slice.append(p)
-        yield "切割执行中", {"__type__": "update", "visible": False}, {"__type__": "update", "visible": True}
+        yield "切割执行中", {"__type__": "update", "visible": False}, {"__type__": "update", "visible": True}, {"__type__": "update"}, {"__type__": "update"}
         for p in ps_slice:
             p.wait()
         ps_slice=[]
-        yield "切割结束",{"__type__":"update","visible":True},{"__type__":"update","visible":False}
+        yield "切割结束", {"__type__":"update","visible":True}, {"__type__":"update","visible":False}, {"__type__": "update", "value":opt_root}, {"__type__": "update", "value":opt_root}
     else:
-        yield "已有正在进行的切割任务，需先终止才能开启下一次任务", {"__type__": "update", "visible": False}, {"__type__": "update", "visible": True}
+        yield "已有正在进行的切割任务，需先终止才能开启下一次任务", {"__type__": "update", "visible": False}, {"__type__": "update", "visible": True}, {"__type__": "update"}, {"__type__": "update"}
 
 def close_slice():
     global ps_slice
@@ -731,25 +734,53 @@ with gr.Blocks(title="GPT-SoVITS WebUI") as app:
                     with gr.Row():
                         asr_model = gr.Dropdown(
                             label       = i18n("ASR 模型"),
-                            choices     = ['SenseVoice'],
+                            choices     = list(asr_dict.keys()),
                             interactive = True,
-                            value="SenseVoice"
+                            value="达摩 ASR (中文)"
                         )
                         asr_size = gr.Dropdown(
                             label       = i18n("ASR 模型尺寸"),
-                            choices     = ["small"],
+                            choices     = ["large"],
                             interactive = True,
-                            value="small"
+                            value="large"
                         )
                         asr_lang = gr.Dropdown(
                             label       = i18n("ASR 语言设置"),
-                            choices     = ["auto","zh","en","ja"],
+                            choices     = ["zh"],
                             interactive = True,
-                            value="auto"
+                            value="zh"
+                        )
+                        asr_precision = gr.Dropdown(
+                            label       = i18n("ASR 语言设置"),
+                            choices     = ["zh"],
+                            interactive = True,
+                            value="zh"
                         )
                     with gr.Row():
                         asr_info = gr.Textbox(label=i18n("ASR进程输出信息"))
 
+                def change_lang_choices(key): #根据选择的模型修改可选的语言
+                    # return gr.Dropdown(choices=asr_dict[key]['lang'])
+                    return {"__type__": "update", "choices": asr_dict[key]['lang'],"value":asr_dict[key]['lang'][0]}
+                def change_size_choices(key): # 根据选择的模型修改可选的模型尺寸
+                    # return gr.Dropdown(choices=asr_dict[key]['size'])
+                    return {"__type__": "update", "choices": asr_dict[key]['size'],"value":asr_dict[key]['size'][-1]}
+                def change_precision_choices(key): #根据选择的模型修改可选的语言
+                    if key =="Faster Whisper (多语种)":
+                        if default_batch_size <= 4:
+                            precision = 'int8'
+                        elif is_half:
+                            precision = 'float16'
+                        else:
+                            precision = 'float32'
+                    else:
+                        precision = 'float32'
+                    # return gr.Dropdown(choices=asr_dict[key]['lang'])
+                    return {"__type__": "update", "choices": asr_dict[key]['precision'],"value":precision}
+                asr_model.change(change_lang_choices, [asr_model], [asr_lang])
+                asr_model.change(change_size_choices, [asr_model], [asr_size])
+                asr_model.change(change_size_choices, [asr_model], [asr_precision])
+
                 
             gr.Markdown(value=i18n("0d-语音文本校对标注工具"))
             with gr.Row():
@@ -762,11 +793,11 @@ with gr.Blocks(title="GPT-SoVITS WebUI") as app:
                 label_info = gr.Textbox(label=i18n("打标工具进程输出信息"))
             if_label.change(change_label, [if_label,path_list], [label_info])
             if_uvr5.change(change_uvr5, [if_uvr5], [uvr5_info])
-            open_asr_button.click(open_asr, [asr_inp_dir, asr_opt_dir, asr_model, asr_size, asr_lang], [asr_info,open_asr_button,close_asr_button,path_list])
+            open_asr_button.click(open_asr, [asr_inp_dir, asr_opt_dir, asr_model, asr_size, asr_lang, asr_precision], [asr_info,open_asr_button,close_asr_button,path_list])
             close_asr_button.click(close_asr, [], [asr_info,open_asr_button,close_asr_button])
-            open_slicer_button.click(open_slice, [slice_inp_path,slice_opt_root,threshold,min_length,min_interval,hop_size,max_sil_kept,_max,alpha,n_process], [slicer_info,open_slicer_button,close_slicer_button])
+            open_slicer_button.click(open_slice, [slice_inp_path,slice_opt_root,threshold,min_length,min_interval,hop_size,max_sil_kept,_max,alpha,n_process], [slicer_info,open_slicer_button,close_slicer_button,asr_inp_dir,denoise_input_dir])
             close_slicer_button.click(close_slice, [], [slicer_info,open_slicer_button,close_slicer_button])
-            open_denoise_button.click(open_denoise, [denoise_input_dir,denoise_output_dir], [denoise_info,open_denoise_button,close_denoise_button])
+            open_denoise_button.click(open_denoise, [denoise_input_dir,denoise_output_dir], [denoise_info,open_denoise_button,close_denoise_button,asr_inp_dir])
             close_denoise_button.click(close_denoise, [], [denoise_info,open_denoise_button,close_denoise_button])
 
         with gr.TabItem(i18n("1-GPT-SoVITS-TTS")):