优化tts逻辑

.gitignore

@@ -71,9 +71,7 @@ examples/
 repositories/
 
 # Workflows - ignore user customizations but keep defaults
-workflows/*
+
-!workflows/*_default.json
-!workflows/README.md
 
 # Templates - ignore user customizations but keep presets
 templates/*

README.md

@@ -147,7 +147,7 @@ uv run streamlit run web/app.py
 
 **ComfyUI 工作流**  
 - 选择图像生成的工作流文件
-- 默认使用 `image_default.json`
+- 默认使用 `image_flux.json`
 - 如果懂 ComfyUI，可以放自己的工作流到 `workflows/` 文件夹
 
 **提示词前缀（Prompt Prefix）**  

config.example.yaml

@@ -17,22 +17,22 @@ llm:
 # DeepSeek:        base_url: "https://api.deepseek.com"                           model: "deepseek-chat"
 # Ollama (Local):  base_url: "http://localhost:11434/v1"                          model: "llama3.2"
 
-# ==================== TTS Configuration ====================
+# ==================== ComfyUI Configuration ====================
-tts:
+comfyui:
-  default: edge  # "edge" (free) or "tts_xxx.json" (ComfyUI workflow)
+  # Global ComfyUI settings
+  comfyui_url: http://127.0.0.1:8188  # ComfyUI server URL (required for selfhost workflows)
+  runninghub_api_key: ""  # RunningHub API key (required for runninghub workflows)
   
-# ==================== Image Generation Configuration ====================
+  # TTS-specific configuration
-image:
+  tts:
+    default: selfhost/tts_edge.json  # TTS workflow to use
+  
+  # Image-specific configuration
+  image:
     # Required: Default workflow to use (no fallback)
-  # Options: runninghub/image_default.json (recommended, no local setup)
+    # Options: runninghub/image_flux.json (recommended, no local setup)
-  #          selfhost/image_default.json (requires local ComfyUI)
+    #          selfhost/image_flux.json (requires local ComfyUI)
-  default_workflow: runninghub/image_default.json
+    default_workflow: runninghub/image_flux.json
-  
-  # Local ComfyUI configuration (required if using selfhost workflows)
-  comfyui_url: http://127.0.0.1:8188
-  
-  # RunningHub cloud configuration (required if using runninghub workflows)
-  runninghub_api_key: ""
     
     # Image prompt prefix (optional)
     prompt_prefix: "Pure white background, minimalist illustration, matchstick figure style, black and white line drawing, simple clean lines"

reelforge/config/__init__.py

@@ -17,7 +17,7 @@ Usage:
     if config_manager.validate():
         print("Config is valid!")
 """
-from .schema import ReelForgeConfig, LLMConfig, TTSConfig, ImageConfig
+from .schema import ReelForgeConfig, LLMConfig, ComfyUIConfig, TTSSubConfig, ImageSubConfig
 from .manager import ConfigManager
 from .loader import load_config_dict, save_config_dict
 
@@ -27,8 +27,9 @@ config_manager = ConfigManager()
 __all__ = [
     "ReelForgeConfig",
     "LLMConfig", 
-    "TTSConfig",
+    "ComfyUIConfig",
-    "ImageConfig",
+    "TTSSubConfig",
+    "ImageSubConfig",
     "ConfigManager",
     "config_manager",
     "load_config_dict",

reelforge/config/manager.py

@@ -93,21 +93,26 @@ class ConfigManager:
             }
         })
     
-    def get_image_config(self) -> dict:
+    def get_comfyui_config(self) -> dict:
-        """Get image configuration as dict"""
+        """Get ComfyUI configuration as dict"""
         return {
-            "default_workflow": self.config.image.default_workflow,
+            "comfyui_url": self.config.comfyui.comfyui_url,
-            "comfyui_url": self.config.image.comfyui_url,
+            "runninghub_api_key": self.config.comfyui.runninghub_api_key,
-            "runninghub_api_key": self.config.image.runninghub_api_key,
+            "tts": {
-            "prompt_prefix": self.config.image.prompt_prefix,
+                "default_workflow": self.config.comfyui.tts.default_workflow,
+            },
+            "image": {
+                "default_workflow": self.config.comfyui.image.default_workflow,
+                "prompt_prefix": self.config.comfyui.image.prompt_prefix,
+            }
         }
     
-    def set_image_config(
+    def set_comfyui_config(
         self, 
         comfyui_url: Optional[str] = None, 
         runninghub_api_key: Optional[str] = None
     ):
-        """Set image configuration"""
+        """Set ComfyUI global configuration"""
         updates = {}
         if comfyui_url is not None:
             updates["comfyui_url"] = comfyui_url
@@ -115,5 +120,5 @@ class ConfigManager:
             updates["runninghub_api_key"] = runninghub_api_key
         
         if updates:
-            self.update({"image": updates})
+            self.update({"comfyui": updates})
 

reelforge/config/schema.py

@@ -13,32 +13,37 @@ class LLMConfig(BaseModel):
     model: str = Field(default="", description="LLM Model Name")
 
 
-class TTSConfig(BaseModel):
+class TTSSubConfig(BaseModel):
-    """TTS configuration"""
+    """TTS-specific configuration (under comfyui.tts)"""
     model_config = {"populate_by_name": True}  # Allow both field name and alias
     
-    default_workflow: str = Field(default="edge", description="Default TTS workflow", alias="default")
+    default_workflow: str = Field(default=None, description="Default TTS workflow (required, no fallback)", alias="default")
 
 
-class ImageConfig(BaseModel):
+class ImageSubConfig(BaseModel):
-    """Image generation configuration"""
+    """Image-specific configuration (under comfyui.image)"""
     model_config = {"populate_by_name": True}  # Allow both field name and alias
     
     default_workflow: str = Field(default=None, description="Default image workflow (required, no fallback)", alias="default")
-    comfyui_url: str = Field(default="http://127.0.0.1:8188", description="ComfyUI Server URL")
-    runninghub_api_key: str = Field(default="", description="RunningHub API Key (optional)")
     prompt_prefix: str = Field(
         default="Pure white background, minimalist illustration, matchstick figure style, black and white line drawing, simple clean lines",
         description="Prompt prefix for all image generation"
     )
 
 
+class ComfyUIConfig(BaseModel):
+    """ComfyUI configuration (includes global settings and service-specific configs)"""
+    comfyui_url: str = Field(default="http://127.0.0.1:8188", description="ComfyUI Server URL")
+    runninghub_api_key: str = Field(default="", description="RunningHub API Key (optional)")
+    tts: TTSSubConfig = Field(default_factory=TTSSubConfig, description="TTS-specific configuration")
+    image: ImageSubConfig = Field(default_factory=ImageSubConfig, description="Image-specific configuration")
+
+
 class ReelForgeConfig(BaseModel):
     """ReelForge main configuration"""
     project_name: str = Field(default="ReelForge", description="Project name")
     llm: LLMConfig = Field(default_factory=LLMConfig)
-    tts: TTSConfig = Field(default_factory=TTSConfig)
+    comfyui: ComfyUIConfig = Field(default_factory=ComfyUIConfig)
-    image: ImageConfig = Field(default_factory=ImageConfig)
     
     def is_llm_configured(self) -> bool:
         """Check if LLM is properly configured"""

reelforge/services/comfy_base_service.py

@@ -19,7 +19,7 @@ class ComfyBaseService:
     
     Subclasses should define:
     - WORKFLOW_PREFIX: Prefix for workflow files (e.g., "image_", "tts_")
-    - DEFAULT_WORKFLOW: Default workflow filename (e.g., "image_default.json")
+    - DEFAULT_WORKFLOW: Default workflow filename (e.g., "image_flux.json")
     - WORKFLOWS_DIR: Directory containing workflows (default: "workflows")
     """
     
@@ -35,7 +35,13 @@ class ComfyBaseService:
             config: Full application config dict
             service_name: Service name in config (e.g., "tts", "image")
         """
-        self.config = config.get(service_name, {})
+        # Service-specific config (e.g., config["comfyui"]["tts"])
+        comfyui_config = config.get("comfyui", {})
+        self.config = comfyui_config.get(service_name, {})
+        
+        # Global ComfyUI config (for comfyui_url and runninghub_api_key)
+        self.global_config = comfyui_config
+        
         self.service_name = service_name
         self._workflows_cache: Optional[List[str]] = None
     
@@ -47,18 +53,18 @@ class ComfyBaseService:
             List of workflow info dicts
             Example: [
                 {
-                    "name": "image_default.json",
+                    "name": "image_flux.json",
-                    "display_name": "image_default.json - Selfhost",
+                    "display_name": "image_flux.json - Selfhost",
                     "source": "selfhost",
-                    "path": "workflows/selfhost/image_default.json",
+                    "path": "workflows/selfhost/image_flux.json",
-                    "key": "selfhost/image_default.json"
+                    "key": "selfhost/image_flux.json"
                 },
                 {
-                    "name": "image_default.json",
+                    "name": "image_flux.json",
-                    "display_name": "image_default.json - Runninghub", 
+                    "display_name": "image_flux.json - Runninghub", 
                     "source": "runninghub",
-                    "path": "workflows/runninghub/image_default.json",
+                    "path": "workflows/runninghub/image_flux.json",
-                    "key": "runninghub/image_default.json",
+                    "key": "runninghub/image_flux.json",
                     "workflow_id": "123456"
                 }
             ]
@@ -101,11 +107,11 @@ class ComfyBaseService:
         Returns:
             Workflow info dict with structure:
             {
-                "name": "image_default.json",
+                "name": "image_flux.json",
-                "display_name": "image_default.json - Runninghub",
+                "display_name": "image_flux.json - Runninghub",
                 "source": "runninghub",
-                "path": "workflows/runninghub/image_default.json",
+                "path": "workflows/runninghub/image_flux.json",
-                "key": "runninghub/image_default.json",
+                "key": "runninghub/image_flux.json",
                 "workflow_id": "123456"  # Only for RunningHub
             }
         """
@@ -134,7 +140,7 @@ class ComfyBaseService:
         Get default workflow from config (required, no fallback)
         
         Returns:
-            Default workflow key (e.g., "runninghub/image_default.json")
+            Default workflow key (e.g., "runninghub/image_flux.json")
         
         Raises:
             ValueError: If default_workflow not configured
@@ -155,17 +161,17 @@ class ComfyBaseService:
         Resolve workflow key to workflow info
         
         Args:
-            workflow: Workflow key (e.g., "runninghub/image_default.json")
+            workflow: Workflow key (e.g., "runninghub/image_flux.json")
                      If None, uses default from config
         
         Returns:
             Workflow info dict with structure:
             {
-                "name": "image_default.json",
+                "name": "image_flux.json",
-                "display_name": "image_default.json - Runninghub",
+                "display_name": "image_flux.json - Runninghub",
                 "source": "runninghub",
-                "path": "workflows/runninghub/image_default.json",
+                "path": "workflows/runninghub/image_flux.json",
-                "key": "runninghub/image_default.json",
+                "key": "runninghub/image_flux.json",
                 "workflow_id": "123456"  # Only for RunningHub
             }
         
@@ -210,19 +216,19 @@ class ComfyBaseService:
         """
         kit_config = {}
         
-        # ComfyUI URL (priority: param > config > env > default)
+        # ComfyUI URL (priority: param > global config > env > default)
         final_comfyui_url = (
             comfyui_url 
-            or self.config.get("comfyui_url")
+            or self.global_config.get("comfyui_url")
             or os.getenv("COMFYUI_BASE_URL")
             or "http://127.0.0.1:8188"
         )
         kit_config["comfyui_url"] = final_comfyui_url
         
-        # RunningHub API key (priority: param > config > env)
+        # RunningHub API key (priority: param > global config > env)
         final_rh_key = (
             runninghub_api_key
-            or self.config.get("runninghub_api_key")
+            or self.global_config.get("runninghub_api_key")
             or os.getenv("RUNNINGHUB_API_KEY")
         )
         if final_rh_key:
@@ -242,11 +248,11 @@ class ComfyBaseService:
             workflows = service.list_workflows()
             # [
             #     {
-            #         "name": "image_default.json",
+            #         "name": "image_flux.json",
-            #         "display_name": "image_default.json - Runninghub",
+            #         "display_name": "image_flux.json - Runninghub",
             #         "source": "runninghub",
-            #         "path": "workflows/runninghub/image_default.json",
+            #         "path": "workflows/runninghub/image_flux.json",
-            #         "key": "runninghub/image_default.json",
+            #         "key": "runninghub/image_flux.json",
             #         "workflow_id": "123456"
             #     },
             #     ...
@@ -260,7 +266,7 @@ class ComfyBaseService:
         List available workflow keys
         
         Returns:
-            List of available workflow keys (e.g., ["runninghub/image_default.json", ...])
+            List of available workflow keys (e.g., ["runninghub/image_flux.json", ...])
         
         Example:
             print(f"Available workflows: {service.available}")

reelforge/services/image.py

@@ -17,7 +17,7 @@ class ImageService(ComfyBaseService):
     Uses ComfyKit to execute image generation workflows.
     
     Usage:
-        # Use default workflow (workflows/image_default.json)
+        # Use default workflow (workflows/image_flux.json)
         image_url = await reelforge.image(prompt="a cat")
         
         # Use specific workflow
@@ -65,7 +65,7 @@ class ImageService(ComfyBaseService):
         
         Args:
             prompt: Image generation prompt
-            workflow: Workflow filename (default: from config or "image_default.json")
+            workflow: Workflow filename (default: from config or "image_flux.json")
             comfyui_url: ComfyUI URL (optional, overrides config)
             runninghub_api_key: RunningHub API key (optional, overrides config)
             width: Image width
@@ -81,7 +81,7 @@ class ImageService(ComfyBaseService):
             Generated image URL/path
         
         Examples:
-            # Simplest: use default workflow (workflows/image_default.json)
+            # Simplest: use default workflow (workflows/image_flux.json)
             image_url = await reelforge.image(prompt="a beautiful cat")
             
             # Use specific workflow

reelforge/services/tts_service.py

@@ -1,39 +1,29 @@
 """
-TTS (Text-to-Speech) Service - Dual implementation (Edge TTS + ComfyUI)
+TTS (Text-to-Speech) Service - ComfyUI Workflow-based implementation
 """
 
-import uuid
 from typing import Optional
 
 from comfykit import ComfyKit
 from loguru import logger
 
 from reelforge.services.comfy_base_service import ComfyBaseService
-from reelforge.utils.os_util import get_temp_path
 
 
 class TTSService(ComfyBaseService):
     """
-    TTS (Text-to-Speech) service - Dual implementation
+    TTS (Text-to-Speech) service - Workflow-based
     
-    Supports two TTS methods:
+    Uses ComfyKit to execute TTS workflows.
-    1. Edge TTS (default) - Free, local SDK, no workflow needed
-    2. ComfyUI Workflow - Workflow-based, requires ComfyUI setup
     
     Usage:
-        # Use default (edge-tts)
+        # Use default workflow
         audio_path = await reelforge.tts(text="Hello, world!")
         
-        # Explicitly use edge-tts
+        # Use specific workflow
         audio_path = await reelforge.tts(
             text="你好，世界！",
-            workflow="edge"
+            workflow="tts_edge.json"
-        )
-        
-        # Use ComfyUI workflow
-        audio_path = await reelforge.tts(
-            text="Hello",
-            workflow="tts_comfyui.json"
         )
         
         # List available workflows
@@ -41,12 +31,9 @@ class TTSService(ComfyBaseService):
     """
     
     WORKFLOW_PREFIX = "tts_"
-    DEFAULT_WORKFLOW = "edge"  # Default to edge-tts
+    DEFAULT_WORKFLOW = None  # No hardcoded default, must be configured
     WORKFLOWS_DIR = "workflows"
     
-    # Built-in providers (not workflow files)
-    BUILTIN_PROVIDERS = ["edge", "edge-tts"]
-    
     def __init__(self, config: dict):
         """
         Initialize TTS service
@@ -56,81 +43,53 @@ class TTSService(ComfyBaseService):
         """
         super().__init__(config, service_name="tts")
     
-    def _resolve_workflow(self, workflow: Optional[str] = None) -> str:
-        """
-        Resolve workflow to actual workflow path or provider name
-        
-        Args:
-            workflow: Workflow filename or provider name (e.g., "edge", "tts_default.json")
-        
-        Returns:
-            Workflow file path or provider name
-        """
-        # 1. If not specified, use default
-        if workflow is None:
-            workflow = self._get_default_workflow()
-        
-        # 2. If it's a built-in provider, return as-is
-        if workflow in self.BUILTIN_PROVIDERS:
-            logger.debug(f"Using built-in TTS provider: {workflow}")
-            return workflow
-        
-        # 3. Otherwise, treat as workflow file (use parent logic)
-        return super()._resolve_workflow(workflow)
     
     async def __call__(
         self,
         text: str,
         workflow: Optional[str] = None,
-        # ComfyUI connection (optional overrides, only for workflow mode)
+        # ComfyUI connection (optional overrides)
         comfyui_url: Optional[str] = None,
         runninghub_api_key: Optional[str] = None,
-        # Common TTS parameters (work for both edge-tts and workflows)
+        # TTS parameters
         voice: Optional[str] = None,
-        rate: Optional[str] = None,
+        speed: float = 1.0,
-        volume: Optional[str] = None,
-        pitch: Optional[str] = None,
         # Output path
         output_path: Optional[str] = None,
         **params
     ) -> str:
         """
-        Generate speech using edge-tts or ComfyUI workflow
+        Generate speech using ComfyUI workflow
         
         Args:
             text: Text to convert to speech
-            workflow: Workflow filename or provider name (default: "edge")
+            workflow: Workflow filename (default: from config)
-                     - "edge" or "edge-tts": Use local edge-tts SDK
+            comfyui_url: ComfyUI URL (optional, overrides config)
-                     - "tts_xxx.json": Use ComfyUI workflow
+            runninghub_api_key: RunningHub API key (optional, overrides config)
-                     - Absolute path/URL/RunningHub ID: Also supported
+            voice: Voice ID (workflow-specific)
-            comfyui_url: ComfyUI URL (only for workflow mode)
+            speed: Speech speed multiplier (1.0 = normal, >1.0 = faster, <1.0 = slower)
-            runninghub_api_key: RunningHub API key (only for workflow mode)
-            voice: Voice ID
-            rate: Speech rate (e.g., "+0%", "+50%", "-20%")
-            volume: Speech volume (e.g., "+0%")
-            pitch: Speech pitch (e.g., "+0Hz")
             output_path: Custom output path (auto-generated if None)
-            **params: Additional parameters
+            **params: Additional workflow parameters
         
         Returns:
             Generated audio file path
         
         Examples:
-            # Simplest: use default (edge-tts)
+            # Simplest: use default workflow
             audio_path = await reelforge.tts(text="Hello, world!")
             
-            # Explicitly use edge-tts with parameters
+            # Use specific workflow
             audio_path = await reelforge.tts(
                 text="你好，世界！",
-                workflow="edge",
+                workflow="tts_edge.json"
-                voice="zh-CN-XiaoxiaoNeural",
-                rate="+20%"
             )
             
-            # Use ComfyUI workflow
+            # With voice and speed
             audio_path = await reelforge.tts(
                 text="Hello",
-                workflow="tts_default.json"
+                workflow="tts_edge.json",
+                voice="zh-CN-XiaoxiaoNeural",
+                speed=1.2
             )
             
             # With absolute path
@@ -138,92 +97,28 @@ class TTSService(ComfyBaseService):
                 text="Hello",
                 workflow="/path/to/custom_tts.json"
             )
-        """
-        # 1. Check if it's a builtin provider (edge-tts)
-        if workflow in self.BUILTIN_PROVIDERS or workflow is None and self._get_default_workflow() in self.BUILTIN_PROVIDERS:
-            # Use edge-tts
-            return await self._call_edge_tts(
-                text=text,
-                voice=voice,
-                rate=rate,
-                volume=volume,
-                pitch=pitch,
-                output_path=output_path,
-                **params
-            )
             
-        # 2. Use ComfyUI workflow - resolve to structured info
+            # With custom ComfyUI server
+            audio_path = await reelforge.tts(
+                text="Hello",
+                comfyui_url="http://192.168.1.100:8188"
+            )
+        """
+        # 1. Resolve workflow (returns structured info)
         workflow_info = self._resolve_workflow(workflow=workflow)
         
+        # 2. Execute ComfyUI workflow
         return await self._call_comfyui_workflow(
             workflow_info=workflow_info,
             text=text,
             comfyui_url=comfyui_url,
             runninghub_api_key=runninghub_api_key,
             voice=voice,
-            rate=rate,
+            speed=speed,
-            volume=volume,
-            pitch=pitch,
             output_path=output_path,
             **params
         )
     
-    async def _call_edge_tts(
-        self,
-        text: str,
-        voice: Optional[str] = None,
-        rate: Optional[str] = None,
-        volume: Optional[str] = None,
-        pitch: Optional[str] = None,
-        output_path: Optional[str] = None,
-        **params
-    ) -> str:
-        """
-        Generate speech using edge-tts SDK
-        
-        Args:
-            text: Text to convert to speech
-            voice: Voice ID (default: zh-CN-YunjianNeural)
-            rate: Speech rate (default: +0%)
-            volume: Speech volume (default: +0%)
-            pitch: Speech pitch (default: +0Hz)
-            output_path: Custom output path (auto-generated if None)
-            **params: Additional parameters (e.g., retry_count, retry_delay)
-        
-        Returns:
-            Generated audio file path
-        """
-        from reelforge.utils.tts_util import edge_tts
-        
-        logger.info(f"🎙️  Using edge-tts (local SDK)")
-        
-        # Generate output path (use provided path or auto-generate)
-        if output_path is None:
-            output_path = get_temp_path(f"{uuid.uuid4().hex}.mp3")
-        else:
-            # Ensure parent directory exists
-            import os
-            os.makedirs(os.path.dirname(output_path), exist_ok=True)
-        
-        # Call edge-tts with output_path to save directly
-        try:
-            audio_bytes = await edge_tts(
-                text=text,
-                voice=voice or "zh-CN-YunjianNeural",
-                rate=rate or "+0%",
-                volume=volume or "+0%",
-                pitch=pitch or "+0Hz",
-                output_path=output_path,
-                **params
-            )
-            
-            logger.info(f"✅ Generated audio (edge-tts): {output_path}")
-            return output_path
-        
-        except Exception as e:
-            logger.error(f"Edge TTS generation error: {e}")
-            raise
-    
     async def _call_comfyui_workflow(
         self,
         workflow_info: dict,
@@ -231,9 +126,7 @@ class TTSService(ComfyBaseService):
         comfyui_url: Optional[str] = None,
         runninghub_api_key: Optional[str] = None,
         voice: Optional[str] = None,
-        rate: Optional[str] = None,
+        speed: float = 1.0,
-        volume: Optional[str] = None,
-        pitch: Optional[str] = None,
         output_path: Optional[str] = None,
         **params
     ) -> str:
@@ -246,9 +139,7 @@ class TTSService(ComfyBaseService):
             comfyui_url: ComfyUI URL
             runninghub_api_key: RunningHub API key
             voice: Voice ID (workflow-specific)
-            rate: Speech rate (workflow-specific)
+            speed: Speech speed multiplier (workflow-specific)
-            volume: Speech volume (workflow-specific)
-            pitch: Speech pitch (workflow-specific)
             output_path: Custom output path (downloads if URL returned)
             **params: Additional workflow parameters
         
@@ -269,12 +160,8 @@ class TTSService(ComfyBaseService):
         # Add optional TTS parameters
         if voice is not None:
             workflow_params["voice"] = voice
-        if rate is not None:
+        if speed != 1.0:
-            workflow_params["rate"] = rate
+            workflow_params["speed"] = speed
-        if volume is not None:
-            workflow_params["volume"] = volume
-        if pitch is not None:
-            workflow_params["pitch"] = pitch
         
         # Add any additional parameters
         workflow_params.update(params)

web/app.py

@@ -100,11 +100,11 @@ def render_advanced_settings():
     
     # Expand if not configured, collapse if configured
     with st.expander(tr("settings.title"), expanded=not is_configured):
-        # 2-column layout: LLM | Image
+        # 2-column layout: LLM | ComfyUI
-        llm_col, image_col = st.columns(2)
+        llm_col, comfyui_col = st.columns(2)
         
         # ====================================================================
-        # Column 1: LLM Settings (Simplified 3-field format)
+        # Column 1: LLM Settings
         # ====================================================================
         with llm_col:
             with st.container(border=True):
@@ -195,21 +195,21 @@ def render_advanced_settings():
                 )
         
         # ====================================================================
-        # Column 2: Image Settings
+        # Column 2: ComfyUI Settings
         # ====================================================================
-        with image_col:
+        with comfyui_col:
             with st.container(border=True):
-                st.markdown(f"**{tr('settings.image.title')}**")
+                st.markdown(f"**{tr('settings.comfyui.title')}**")
                 
                 # Get current configuration
-                image_config = config_manager.get_image_config()
+                comfyui_config = config_manager.get_comfyui_config()
                 
                 # Local/Self-hosted ComfyUI configuration
-                st.markdown(f"**{tr('settings.image.local_title')}**")
+                st.markdown(f"**{tr('settings.comfyui.local_title')}**")
                 comfyui_url = st.text_input(
-                    tr("settings.image.comfyui_url"),
+                    tr("settings.comfyui.comfyui_url"),
-                    value=image_config.get("comfyui_url", "http://127.0.0.1:8188"),
+                    value=comfyui_config.get("comfyui_url", "http://127.0.0.1:8188"),
-                    help=tr("settings.image.comfyui_url_help"),
+                    help=tr("settings.comfyui.comfyui_url_help"),
                     key="comfyui_url_input"
                 )
                 
@@ -228,12 +228,12 @@ def render_advanced_settings():
                 st.markdown("---")
                 
                 # RunningHub cloud configuration
-                st.markdown(f"**{tr('settings.image.cloud_title')}**")
+                st.markdown(f"**{tr('settings.comfyui.cloud_title')}**")
                 runninghub_api_key = st.text_input(
-                    tr("settings.image.runninghub_api_key"),
+                    tr("settings.comfyui.runninghub_api_key"),
-                    value=image_config.get("runninghub_api_key", ""),
+                    value=comfyui_config.get("runninghub_api_key", ""),
                     type="password",
-                    help=tr("settings.image.runninghub_api_key_help"),
+                    help=tr("settings.comfyui.runninghub_api_key_help"),
                     key="runninghub_api_key_input"
                 )
         
@@ -250,8 +250,8 @@ def render_advanced_settings():
                     if llm_api_key and llm_base_url and llm_model:
                         config_manager.set_llm_config(llm_api_key, llm_base_url, llm_model)
                     
-                    # Save Image configuration
+                    # Save ComfyUI configuration
-                    config_manager.set_image_config(
+                    config_manager.set_comfyui_config(
                         comfyui_url=comfyui_url if comfyui_url else None,
                         runninghub_api_key=runninghub_api_key if runninghub_api_key else None
                     )
@@ -380,53 +380,12 @@ def main():
                 st.info(tr("video.frames_fixed_mode_hint"))
         
         # ====================================================================
-        # Audio Settings (Voice + BGM)
+        # Audio Settings (BGM + TTS)
         # ====================================================================
         with st.container(border=True):
             st.markdown(f"**{tr('section.audio_settings')}**")
             
-            # Voice selection
+            # Background music (moved to top)
-            st.markdown(f"**{tr('voice.title')}**")
-            voice_id = st.selectbox(
-                "Voice",
-                [
-                    "zh-CN-YunjianNeural",  # 男声-专业
-                    "zh-CN-YunxiNeural",    # 男声-年轻
-                    "zh-CN-XiaoxiaoNeural", # 女声-温柔
-                    "zh-CN-XiaoyiNeural",   # 女声-活力
-                ],
-                format_func=lambda x: {
-                    "zh-CN-YunjianNeural": tr("voice.male_professional"),
-                    "zh-CN-YunxiNeural": tr("voice.male_young"),
-                    "zh-CN-XiaoxiaoNeural": tr("voice.female_gentle"),
-                    "zh-CN-XiaoyiNeural": tr("voice.female_energetic"),
-                }[x],
-                label_visibility="collapsed"
-            )
-            
-            # Voice preview button
-            if st.button(tr("voice.preview"), key="preview_voice", use_container_width=True):
-                with st.spinner(tr("voice.previewing")):
-                    try:
-                        # Generate preview audio
-                        preview_text = "大家好，这是一段测试语音。"
-                        
-                        # Use TTS service to generate audio (auto temp path)
-                        audio_path = run_async(reelforge.tts(
-                            text=preview_text,
-                            voice=voice_id
-                        ))
-                        
-                        # Play the audio
-                        if os.path.exists(audio_path):
-                            st.audio(audio_path, format="audio/mp3")
-                        else:
-                            st.error("Failed to generate preview audio")
-                    except Exception as e:
-                        st.error(tr("voice.preview_failed", error=str(e)))
-                        logger.exception(e)
-            
-            # Background music
             st.markdown(f"**{tr('bgm.title')}**")
             st.caption(tr("bgm.custom_help"))
             
@@ -466,6 +425,78 @@ def main():
             # Use full filename for bgm_path (including extension)
             bgm_path = None if bgm_choice == tr("bgm.none") else bgm_choice
             
+            
+            # TTS Workflow selection
+            st.markdown(f"**{tr('tts.title')}**")
+            st.caption(tr("tts.workflow_help"))
+            
+            # Get available TTS workflows
+            tts_workflows = reelforge.tts.list_workflows()
+            
+            # Build options for selectbox
+            tts_workflow_options = [wf["display_name"] for wf in tts_workflows]
+            tts_workflow_keys = [wf["key"] for wf in tts_workflows]
+            
+            # Default to saved workflow if exists
+            default_tts_index = 0
+            comfyui_config = config_manager.get_comfyui_config()
+            saved_tts_workflow = comfyui_config["tts"]["default_workflow"]
+            if saved_tts_workflow and saved_tts_workflow in tts_workflow_keys:
+                default_tts_index = tts_workflow_keys.index(saved_tts_workflow)
+            
+            tts_workflow_display = st.selectbox(
+                "TTS Workflow",
+                tts_workflow_options if tts_workflow_options else ["No TTS workflows found"],
+                index=default_tts_index,
+                label_visibility="collapsed",
+                key="tts_workflow_select"
+            )
+            
+            # Get the actual workflow key
+            if tts_workflow_options:
+                tts_selected_index = tts_workflow_options.index(tts_workflow_display)
+                tts_workflow_key = tts_workflow_keys[tts_selected_index]
+            else:
+                tts_workflow_key = "selfhost/tts_edge.json"  # fallback
+            
+            # TTS preview expander (similar to image preview)
+            with st.expander(tr("tts.preview_title"), expanded=False):
+                # Preview text input
+                preview_text = st.text_input(
+                    tr("tts.preview_text"),
+                    value="大家好，这是一段测试语音。",
+                    placeholder=tr("tts.preview_text_placeholder"),
+                    key="tts_preview_text"
+                )
+                
+                # Preview button
+                if st.button(tr("tts.preview_button"), key="preview_tts", use_container_width=True):
+                    with st.spinner(tr("tts.previewing")):
+                        try:
+                            # Generate preview audio using selected workflow
+                            audio_path = run_async(reelforge.tts(
+                                text=preview_text,
+                                workflow=tts_workflow_key
+                            ))
+                            
+                            # Play the audio
+                            if audio_path:
+                                st.success(tr("tts.preview_success"))
+                                if os.path.exists(audio_path):
+                                    st.audio(audio_path, format="audio/mp3")
+                                elif audio_path.startswith('http'):
+                                    st.audio(audio_path)
+                                else:
+                                    st.error("Failed to generate preview audio")
+                                
+                                # Show file path
+                                st.caption(f"📁 {audio_path}")
+                            else:
+                                st.error("Failed to generate preview audio")
+                        except Exception as e:
+                            st.error(tr("tts.preview_failed", error=str(e)))
+                            logger.exception(e)
+    
     # ========================================================================
     # Middle Column: Visual Settings (Style & Template)
     # ========================================================================
@@ -484,8 +515,8 @@ def main():
             workflows = reelforge.image.list_workflows()
             
             # Build options for selectbox
-            # Display: "image_default.json - Runninghub"
+            # Display: "image_flux.json - Runninghub"
-            # Value: "runninghub/image_default.json"
+            # Value: "runninghub/image_flux.json"
             workflow_options = [wf["display_name"] for wf in workflows]
             workflow_keys = [wf["key"] for wf in workflows]
             
@@ -493,8 +524,8 @@ def main():
             default_workflow_index = 0
             
             # If user has a saved preference in config, try to match it
-            image_config = config_manager.get_image_config()
+            comfyui_config = config_manager.get_comfyui_config()
-            saved_workflow = image_config.get("default_workflow")
+            saved_workflow = comfyui_config["image"]["default_workflow"]
             if saved_workflow and saved_workflow in workflow_keys:
                 default_workflow_index = workflow_keys.index(saved_workflow)
             
@@ -506,20 +537,19 @@ def main():
                 key="image_workflow_select"
             )
             
-            # Get the actual workflow key (e.g., "runninghub/image_default.json")
+            # Get the actual workflow key (e.g., "runninghub/image_flux.json")
             if workflow_options:
                 workflow_selected_index = workflow_options.index(workflow_display)
                 workflow_key = workflow_keys[workflow_selected_index]
             else:
-                workflow_key = "runninghub/image_default.json"  # fallback
+                workflow_key = "runninghub/image_flux.json"  # fallback
             
             
             # 2. Prompt prefix input
             st.caption(tr("style.prompt_prefix"))
             
             # Get current prompt_prefix from config
-            image_config = config_manager.get_image_config()
+            current_prefix = comfyui_config["image"]["prompt_prefix"]
-            current_prefix = image_config.get("prompt_prefix", "")
             
             # Prompt prefix input (temporary, not saved to config)
             prompt_prefix = st.text_area(
@@ -757,8 +787,8 @@ def main():
                         mode=mode,
                         title=title if title else None,
                         n_scenes=n_scenes,
-                        voice_id=voice_id,
+                        tts_workflow=tts_workflow_key,  # Pass TTS workflow key
-                        image_workflow=workflow_key,  # Pass workflow key (e.g., "runninghub/image_default.json")
+                        image_workflow=workflow_key,  # Pass workflow key (e.g., "runninghub/image_flux.json")
                         frame_template=frame_template,
                         prompt_prefix=prompt_prefix,  # Pass prompt_prefix
                         bgm_path=bgm_path,

web/i18n/locales/en_US.json

@@ -58,7 +58,7 @@
     
     "style.title": "🎨 Image Settings",
     "style.workflow": "ComfyUI Workflow",
-    "style.workflow_help": "💡 Custom: Place image_xxx.json in workflows/ folder",
+    "style.workflow_help": "💡 Custom: Place image_xxx.json in workflows/selfhost/ or workflows/runninghub/ folder",
     "style.prompt_prefix": "Style Prompt Prefix",
     "style.prompt_prefix_placeholder": "Enter style prefix (leave empty for config default)",
     "style.prompt_prefix_help": "This text will be automatically added before all image generation prompts. To permanently change, edit config.yaml",
@@ -163,18 +163,24 @@
     "settings.llm.model": "Model",
     "settings.llm.model_help": "Model name",
     
-    "settings.tts.title": "🎤 Text-to-Speech",
+    "settings.comfyui.title": "🔧 ComfyUI Configuration",
-    "settings.tts.provider": "Provider",
+    "settings.comfyui.local_title": "Local/Self-hosted ComfyUI",
-    "settings.tts.provider_help": "Select TTS service provider",
+    "settings.comfyui.cloud_title": "RunningHub Cloud",
-    "settings.tts.edge_info": "💡 Edge TTS is free and requires no configuration",
+    "settings.comfyui.comfyui_url": "ComfyUI Server URL",
+    "settings.comfyui.comfyui_url_help": "Local or remote ComfyUI server address",
+    "settings.comfyui.runninghub_api_key": "RunningHub API Key",
+    "settings.comfyui.runninghub_api_key_help": "Visit https://runninghub.ai to register and get API Key",
     
-    "settings.image.title": "🎨 Image Generation",
+    "tts.title": "🎤 TTS Workflow",
-    "settings.image.local_title": "Local/Self-hosted ComfyUI",
+    "tts.workflow": "TTS Workflow",
-    "settings.image.cloud_title": "RunningHub Cloud",
+    "tts.workflow_help": "💡 Custom: Place tts_xxx.json in workflows/selfhost/ or workflows/runninghub/ folder",
-    "settings.image.comfyui_url": "ComfyUI Service URL",
+    "tts.preview_title": "🔍 Preview TTS",
-    "settings.image.comfyui_url_help": "Local or remote ComfyUI service URL, default: http://127.0.0.1:8188",
+    "tts.preview_text": "Preview Text",
-    "settings.image.runninghub_api_key": "RunningHub API Key",
+    "tts.preview_text_placeholder": "Enter text to preview...",
-    "settings.image.runninghub_api_key_help": "Visit https://runninghub.ai to register and get API Key",
+    "tts.preview_button": "🔊 Generate Preview",
+    "tts.previewing": "Generating TTS preview...",
+    "tts.preview_success": "✅ Preview generated successfully!",
+    "tts.preview_failed": "❌ Preview failed: {error}",
     
     "settings.book.title": "📚 Book Information",
     "settings.book.provider": "Provider",

web/i18n/locales/zh_CN.json

@@ -58,7 +58,7 @@
     
     "style.title": "🎨 插图设置",
     "style.workflow": "生图工作流",
-    "style.workflow_help": "💡 自定义：将 image_xxx.json 放入 workflows/ 文件夹",
+    "style.workflow_help": "💡 自定义：将 image_xxx.json 放入 workflows/selfhost/ 或 workflows/runninghub/ 文件夹",
     "style.prompt_prefix": "风格提示词前缀",
     "style.prompt_prefix_placeholder": "输入风格前缀（留空则使用配置文件默认值）",
     "style.prompt_prefix_help": "此文本将自动添加到所有图像生成提示词之前。要永久修改，请编辑 config.yaml",
@@ -163,18 +163,24 @@
     "settings.llm.model": "Model",
     "settings.llm.model_help": "模型名称",
     
-    "settings.tts.title": "🎤 语音合成",
+    "settings.comfyui.title": "🔧 ComfyUI 配置",
-    "settings.tts.provider": "服务商",
+    "settings.comfyui.local_title": "本地/自建 ComfyUI",
-    "settings.tts.provider_help": "选择 TTS 服务提供商",
+    "settings.comfyui.cloud_title": "RunningHub 云端",
-    "settings.tts.edge_info": "💡 Edge TTS 是免费的，无需配置",
+    "settings.comfyui.comfyui_url": "ComfyUI 服务器地址",
+    "settings.comfyui.comfyui_url_help": "本地或远程 ComfyUI 服务器地址",
+    "settings.comfyui.runninghub_api_key": "RunningHub API 密钥",
+    "settings.comfyui.runninghub_api_key_help": "访问 https://runninghub.ai 注册并获取 API Key",
     
-    "settings.image.title": "🎨 图像生成",
+    "tts.title": "🎤 TTS 工作流",
-    "settings.image.local_title": "本地/自建 ComfyUI",
+    "tts.workflow": "TTS 工作流",
-    "settings.image.cloud_title": "RunningHub 云端",
+    "tts.workflow_help": "💡 自定义：将 tts_xxx.json 放入 workflows/selfhost/ 或 workflows/runninghub/ 文件夹",
-    "settings.image.comfyui_url": "ComfyUI 服务地址",
+    "tts.preview_title": "🔍 预览 TTS",
-    "settings.image.comfyui_url_help": "本地或远程 ComfyUI 服务地址，默认: http://127.0.0.1:8188",
+    "tts.preview_text": "预览文本",
-    "settings.image.runninghub_api_key": "RunningHub API Key",
+    "tts.preview_text_placeholder": "输入要试听的文本...",
-    "settings.image.runninghub_api_key_help": "访问 https://runninghub.ai 注册并获取 API Key",
+    "tts.preview_button": "🔊 生成预览",
+    "tts.previewing": "正在生成 TTS 预览...",
+    "tts.preview_success": "✅ 预览生成成功！",
+    "tts.preview_failed": "❌ 预览失败：{error}",
     
     "settings.book.title": "📚 书籍信息",
     "settings.book.provider": "服务商",

workflows/runninghub/image_flux.json

@@ -0,0 +1,5 @@
+{
+  "source": "runninghub",
+  "workflow_id": "1983427617984585729"
+}
+

workflows/runninghub/tts_edge.json

@@ -0,0 +1,5 @@
+{
+  "source": "runninghub",
+  "workflow_id": "1983513964837543938"
+}
+

workflows/selfhost/image_flux.json

@@ -0,0 +1,155 @@
+{
+  "29": {
+    "inputs": {
+      "seed": 362283278588365,
+      "steps": 20,
+      "cfg": 1.5,
+      "sampler_name": "euler",
+      "scheduler": "simple",
+      "denoise": 1,
+      "model": [
+        "30",
+        0
+      ],
+      "positive": [
+        "35",
+        0
+      ],
+      "negative": [
+        "33",
+        0
+      ],
+      "latent_image": [
+        "43",
+        0
+      ]
+    },
+    "class_type": "KSampler",
+    "_meta": {
+      "title": "KSampler"
+    }
+  },
+  "30": {
+    "inputs": {
+      "ckpt_name": "flux1-dev-fp8.safetensors"
+    },
+    "class_type": "CheckpointLoaderSimple",
+    "_meta": {
+      "title": "Load Checkpoint"
+    }
+  },
+  "31": {
+    "inputs": {
+      "text": [
+        "46",
+        0
+      ],
+      "clip": [
+        "30",
+        1
+      ]
+    },
+    "class_type": "CLIPTextEncode",
+    "_meta": {
+      "title": "CLIP Text Encode (Prompt)"
+    }
+  },
+  "33": {
+    "inputs": {
+      "conditioning": [
+        "31",
+        0
+      ]
+    },
+    "class_type": "ConditioningZeroOut",
+    "_meta": {
+      "title": "ConditioningZeroOut"
+    }
+  },
+  "35": {
+    "inputs": {
+      "guidance": 3.5,
+      "conditioning": [
+        "31",
+        0
+      ]
+    },
+    "class_type": "FluxGuidance",
+    "_meta": {
+      "title": "FluxGuidance"
+    }
+  },
+  "36": {
+    "inputs": {
+      "filename_prefix": "ComfyUI",
+      "images": [
+        "37",
+        0
+      ]
+    },
+    "class_type": "SaveImage",
+    "_meta": {
+      "title": "Save Image"
+    }
+  },
+  "37": {
+    "inputs": {
+      "samples": [
+        "29",
+        0
+      ],
+      "vae": [
+        "30",
+        2
+      ]
+    },
+    "class_type": "VAEDecode",
+    "_meta": {
+      "title": "VAE Decode"
+    }
+  },
+  "41": {
+    "inputs": {
+      "value": 512
+    },
+    "class_type": "easy int",
+    "_meta": {
+      "title": "$width.value"
+    }
+  },
+  "42": {
+    "inputs": {
+      "value": 512
+    },
+    "class_type": "easy int",
+    "_meta": {
+      "title": "$height.value"
+    }
+  },
+  "43": {
+    "inputs": {
+      "width": [
+        "41",
+        0
+      ],
+      "height": [
+        "42",
+        0
+      ],
+      "batch_size": 1
+    },
+    "class_type": "EmptyLatentImage",
+    "_meta": {
+      "title": "Empty Latent Image"
+    }
+  },
+  "46": {
+    "inputs": {
+      "value": "a dog"
+    },
+    "class_type": "PrimitiveStringMultiline",
+    "_meta": {
+      "title": "$prompt.value!"
+    }
+  }
+}

workflows/selfhost/tts_edge.json

@@ -0,0 +1,78 @@
+{
+  "1": {
+    "inputs": {
+      "text": [
+        "3",
+        0
+      ],
+      "voice": [
+        "5",
+        0
+      ],
+      "speed": [
+        "8",
+        0
+      ],
+      "pitch": 0
+    },
+    "class_type": "EdgeTTS",
+    "_meta": {
+      "title": "Edge TTS 🔊"
+    }
+  },
+  "3": {
+    "inputs": {
+      "value": "床前明月光，疑是地上霜。"
+    },
+    "class_type": "PrimitiveStringMultiline",
+    "_meta": {
+      "title": "$text.value!"
+    }
+  },
+  "4": {
+    "inputs": {
+      "filename_prefix": "audio/ComfyUI",
+      "quality": "V0",
+      "audioUI": "",
+      "audio": [
+        "1",
+        0
+      ]
+    },
+    "class_type": "SaveAudioMP3",
+    "_meta": {
+      "title": "Save Audio (MP3)"
+    }
+  },
+  "5": {
+    "inputs": {
+      "text": "[Chinese] zh-CN Yunjian",
+      "anything": [
+        "7",
+        0
+      ]
+    },
+    "class_type": "easy showAnything",
+    "_meta": {
+      "title": "Show Any"
+    }
+  },
+  "7": {
+    "inputs": {
+      "value": "[Chinese] zh-CN Yunjian"
+    },
+    "class_type": "PrimitiveStringMultiline",
+    "_meta": {
+      "title": "$voice.value"
+    }
+  },
+  "8": {
+    "inputs": {
+      "value": 1
+    },
+    "class_type": "easy float",
+    "_meta": {
+      "title": "$speed.value"
+    }
+  }
+}