New fix

README.md

@@ -104,6 +104,69 @@
    ```bash
     export HF_ENDPOINT=https://hf-mirror.com
    ```
+## api接口使用
+
+启动接口服务
+
+```
+python3 api.py
+```
+
+get请求：
+
+```
+http://localhost:9880/?text=我问她月饼爱吃咸的还是甜的，那天老色鬼说,你身上的月饼，自然是甜过了蜜糖。&seed=2
+```
+
+post请求：
+
+```
+import requests
+import json
+data = json.dumps({"text":"我问她月饼爱吃咸的还是甜的，那天老色鬼说,你身上的月饼，自然是甜过了蜜糖","seed":2})
+url = "http://localhost:9880"
+headers = {"Content-Type":"application/json"}
+r = requests.post(url,data=data,headers=headers)
+audio_data = r.content
+
+with open(f"测试音频.wav","wb") as f:
+   f.write(audio_data)
+```
+
+接口支持音色向量角色参数roleid，参见webui,支持流式参数:streaming=1
+
+## 本地ollama大模型生成角色剧本
+
+下载权重：https://huggingface.co/shenzhi-wang/Mistral-7B-v0.3-Chinese-Chat-4bit
+
+新建template文件：Modelflie_mistral7b0.3.txt
+
+```
+FROM .\Mistral-7B-v0.3-Chinese-Chat-q4_0.gguf
+
+#设置模型温度（值越小回答越严谨，值越大回答越发散）
+PARAMETER temperature 0.8
+
+#设置上下文token尺寸
+PARAMETER num_ctx 4096
+
+TEMPLATE """[INST] {{ if .System }}{{ .System }} {{ end }}{{ .Prompt }} [/INST]
+"""
+
+SYSTEM """你是一名作家，文辞优美，擅长转写话剧和影视剧的剧本"""
+```
+
+导入模型:
+
+```
+ollama create nsfw -f Modelflie_mistral7b0.3.txt
+```
+
+启动服务：
+
+```
+ollama serve
+```
 
 ## 贡献
 

api.py

@@ -22,6 +22,8 @@
 
 from starlette.middleware.cors import CORSMiddleware  #引入 CORS中间件模块
 
+import json
+
 #设置允许访问的域名
 origins = ["*"]  #"*"，即为所有。
 
@@ -59,8 +61,9 @@ def deterministic(seed=0):
 
 class TTS_Request(BaseModel):
     text: str = None
-    seed: int = 2581
+    seed: int = 7750
     speed: int = 3
+    roleid: str = None
     media_type: str = "wav"
     streaming: int = 0
 
@@ -156,7 +159,7 @@ def pack_audio(io_buffer:BytesIO, data:np.ndarray, rate:int, media_type:str):
     return io_buffer
 
 
-def generate_tts_audio(text_file,seed=2581,speed=3, oral=0, laugh=0, bk=4, min_length=10, batch_size=5, temperature=0.3, top_P=0.7,
+def generate_tts_audio(text_file,seed=7750,roleid=None,speed=3, oral=0, laugh=0, bk=4, min_length=10, batch_size=5, temperature=0.3, top_P=0.7,
                        top_K=20,streaming=0,cur_tqdm=None):
 
     from utils import combine_audio, save_audio, batch_split
@@ -178,8 +181,24 @@ def generate_tts_audio(text_file,seed=2581,speed=3, oral=0, laugh=0, bk=4, min_l
     refine_text_prompt = f"[oral_{oral}][laugh_{laugh}][break_{bk}]"
 
 
-    deterministic(seed)
-    rnd_spk_emb = chat.sample_random_speaker()
+
+    if not roleid:
+        deterministic(seed)
+        rnd_spk_emb = chat.sample_random_speaker()
+    else:
+
+        # 从 JSON 文件中读取数据
+        with open('./slct_voice_240605.json', 'r', encoding='utf-8') as json_file:
+            slct_idx_loaded = json.load(json_file)
+
+        # 将包含 Tensor 数据的部分转换回 Tensor 对象
+        for key in slct_idx_loaded:
+            tensor_list = slct_idx_loaded[key]["tensor"]
+            slct_idx_loaded[key]["tensor"] = torch.tensor(tensor_list)
+
+        # 将音色 tensor 打包进params_infer_code，固定使用此音色发音，调低temperature
+        rnd_spk_emb = slct_idx_loaded[roleid]["tensor"]
+
     params_infer_code = {
         'spk_emb': rnd_spk_emb,
         'prompt': f'[speed_{speed}]',
@@ -265,8 +284,10 @@ async def tts_handle(req:dict):
     print(req["streaming"])
 
     if not req["streaming"]:
+
+        print(req["roleid"])
 
-        audio_data = next(generate_tts_audio(req["text"],req["seed"]))
+        audio_data = next(generate_tts_audio(req["text"],req["seed"]),req["roleid"])
 
         # print(audio_data)
 
@@ -282,7 +303,7 @@ async def tts_handle(req:dict):
 
     else:
 
-        tts_generator = generate_tts_audio(req["text"],req["seed"],streaming=1)
+        tts_generator = generate_tts_audio(req["text"],req["seed"],roleid=req["roleid"],streaming=1)
 
         sr = 24000
 
@@ -297,14 +318,15 @@ def streaming_generator(tts_generator:Generator, media_type:str):
 
 
 @app.get("/")
-async def tts_get(text: str = None,media_type:str = "wav",seed:int = 2581,streaming:int = 0):
+async def tts_get(text: str = None,media_type:str = "wav",seed:int = 2581,streaming:int = 0,roleid:str = None):
     req = {
         "text": text,
         "media_type": media_type,
         "seed": seed,
         "streaming": streaming,
+        "roleid": roleid,
     }
-    print("第一次")
+
     return await tts_handle(req)
 
 

llm_utils.py

@@ -9,6 +9,44 @@
 from tqdm import tqdm
 from config import LLM_RETRIES, LLM_REQUEST_INTERVAL, LLM_RETRY_DELAY, LLM_MAX_TEXT_LENGTH, LLM_PROMPT
 
+import requests
+
+
+def ollama_generate(url,model,text):
+    from config import LLM_PROMPT
+
+
+    LLM_PROMPT += f"\n注意，只返回json即可，不要返回其他格式\n{text}"
+
+    data = {
+    "model":model,
+    "prompt":LLM_PROMPT,
+    "stream":False
+    }
+
+    data = json.dumps(data)
+    headers = {"Content-Type":"application/json"}#指定提交的是json
+    r = requests.post(url,data=data,headers=headers)
+
+    try:
+        res = json.loads(r.text)
+
+        clist = json.loads(res["response"].replace("```json","").replace("```",""))
+
+        texts = ""
+
+        for x in clist:
+
+            texts += f"{x['character']}::{x['txt']}\n"
+    except Exception as e:
+
+        print(str(e))
+
+        texts = "报错了，请重新生成 \n"
+
+
+    return texts
+
 
 def send_request(client, prompt, text, model):
     text = remove_json_escape_characters(text)

webui_mix.py

@@ -17,6 +17,8 @@
 from config import DEFAULT_BATCH_SIZE, DEFAULT_SPEED, DEFAULT_TEMPERATURE, DEFAULT_TOP_K, DEFAULT_TOP_P, DEFAULT_ORAL, \
     DEFAULT_LAUGH, DEFAULT_BK, DEFAULT_SEG_LENGTH
 
+from llm_utils import ollama_generate
+
 parser = argparse.ArgumentParser(description="Gradio ChatTTS MIX")
 parser.add_argument("--source", type=str, default="huggingface", help="Model source: 'huggingface' or 'local'.")
 parser.add_argument("--local_path", type=str, help="Path to local model if source is 'local'.")
@@ -55,6 +57,15 @@
 # chat.load_models(source="local", local_path="models")
 # torch.cuda.empty_cache()
 
+def file_show(file):
+    if file is None:
+        return ""
+    try:
+      with open(file.name, "r", encoding="utf-8") as f:
+         text = f.read()
+      return text
+    except Exception as error:
+        return error
 
 # 加载
 def load_seeds():
@@ -435,7 +446,8 @@ def inser_token(text, btn):
                     break_button = gr.Button("+停顿", variant="secondary")
                     laugh_button = gr.Button("+笑声", variant="secondary")
                 refine_button = gr.Button("Refine Text（预处理 加入停顿词、笑声等）", variant="secondary")
-
+                srt_file = gr.File(label="上传文本",file_types=['.txt'],file_count='single')
+            srt_file.change(file_show,inputs=[srt_file],outputs=[text_file_input])
             with gr.Column():
                 gr.Markdown("### 配置参数")
                 with gr.Row():
@@ -587,7 +599,8 @@ def llm_change(model):
                 "gpt-3.5-turbo-0125": ["https://api.openai.com/v1"],
                 "gpt-4o": ["https://api.openai.com/v1"],
                 "deepseek-chat": ["https://api.deepseek.com"],
-                "yi-large": ["https://api.lingyiwanwu.com/v1"]
+                "yi-large": ["https://api.lingyiwanwu.com/v1"],
+                "本地ollama": ["http://localhost:11434/api/generate"]
             }
             if model in llm_setting:
                 return llm_setting[model][0]
@@ -775,20 +788,22 @@ def batch(iterable, batch_size):
 
                 with gr.Row(equal_height=True):
                     # 选择模型 只有 gpt4o deepseek-chat yi-large 三个选项
-                    model_select = gr.Radio(label="选择模型", choices=["gpt-4o", "deepseek-chat", "yi-large"],
+                    model_select = gr.Radio(label="选择模型", choices=["gpt-4o", "deepseek-chat", "yi-large","本地ollama"],
                                             value="gpt-4o", interactive=True, )
                 with gr.Row(equal_height=True):
                     openai_api_base_input = gr.Textbox(label="OpenAI API Base URL",
                                                        placeholder="请输入API Base URL",
                                                        value=r"https://api.openai.com/v1")
                     openai_api_key_input = gr.Textbox(label="OpenAI API Key", placeholder="请输入API Key",
                                                       value="sk-xxxxxxx",type="password")
+                    ollama_model = gr.Textbox(label="ollama本地模型名称",value="nsfw")
                 # AI提示词
                 ai_text_input = gr.Textbox(label="剧情简介或者一段故事", placeholder="请输入文本...", lines=2,
                                            value=ai_text_default)
 
                 # 生成脚本的按钮
                 ai_script_generate_button = gr.Button("AI脚本生成")
+                ai_script_generate_button_ollama = gr.Button("本地Ollama模型AI脚本生成")
 
             with gr.Column(scale=3):
                 gr.Markdown("### 脚本")
@@ -844,6 +859,12 @@ def batch(iterable, batch_size):
             inputs=[model_select, openai_api_base_input, openai_api_key_input, ai_text_input],
             outputs=[script_text_input]
         )
+        # AI脚本生成 ollama
+        ai_script_generate_button_ollama.click(
+            ollama_generate,
+            inputs=[openai_api_base_input,ollama_model,ai_text_input],
+            outputs=[script_text_input]
+        )
         # 音频生成
         script_generate_audio.click(
             generate_script_audio,