let5see/AI-Video
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

更新TTS相关配置，调整语音ID格式，优化工作流参数，确保一致性和可读性。

Browse Source
This commit is contained in:
puke
2025-10-30 00:06:23 +08:00
parent fb18adf318
commit f7ad45354e
12 changed files with 43 additions and 37 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
reelforge/services/frame_processor.py
View File

@@ -124,11 +124,12 @@ class FrameProcessor:
from reelforge.utils.os_util import get_task_frame_path
output_path = get_task_frame_path(config.task_id, frame.index, "audio")
# Call TTS with specific output path
# Call TTS with specific output path and workflow
audio_path = await self.core.tts(
text=frame.narration,
workflow=config.tts_workflow, # Use workflow from config
voice=config.voice_id,
rate="+20%",
speed=config.tts_speed, # Use speed (not rate) from config
output_path=output_path,
)

4
reelforge/services/image_prompt_generator.py
View File

@@ -116,8 +116,8 @@ class ImagePromptGeneratorService:
# 5. Apply prompt prefix to each prompt
from reelforge.utils.prompt_helper import build_image_prompt
# Get prompt prefix from config
image_config = self.core.config.get("image", {})
# Get prompt prefix from config (fix: correct path is comfyui.image.prompt_prefix)
image_config = self.core.config.get("comfyui", {}).get("image", {})
prompt_prefix = image_config.get("prompt_prefix", "")
# Apply prefix to each base prompt

6
reelforge/services/tts_service.py
View File

@@ -52,8 +52,8 @@ class TTSService(ComfyBaseService):
comfyui_url: Optional[str] = None,
runninghub_api_key: Optional[str] = None,
# TTS parameters
voice: Optional[str] = None,
speed: float = 1.0,
voice: str = "[Chinese] zh-CN Yunjian",
speed: float = 1.2,
# Output path
output_path: Optional[str] = None,
**params
@@ -88,7 +88,7 @@ class TTSService(ComfyBaseService):
audio_path = await reelforge.tts(
text="Hello",
workflow="tts_edge.json",
voice="zh-CN-XiaoxiaoNeural",
voice="[Chinese] zh-CN Xiaoxiao",
speed=1.2
)

13
reelforge/services/video_generator.py
View File

@@ -54,7 +54,9 @@ class VideoGeneratorService:
# === Basic Config ===
n_scenes: int = 5, # Only used in generate mode; ignored in fixed mode
voice_id: str = "zh-CN-YunjianNeural",
voice_id: str = "[Chinese] zh-CN Yunjian",
tts_workflow: Optional[str] = None,
tts_speed: float = 1.2,
output_path: Optional[str] = None,
# === LLM Parameters ===
@@ -111,7 +113,9 @@ class VideoGeneratorService:
n_scenes: Number of storyboard scenes (default 5)
Only effective in generate mode; ignored in fixed mode
voice_id: TTS voice ID (default "zh-CN-YunjianNeural")
voice_id: TTS voice ID (default "[Chinese] zh-CN Yunjian")
tts_workflow: TTS workflow filename (e.g., "tts_edge.json", None = use default)
tts_speed: TTS speed multiplier (1.0 = normal, 1.2 = 20% faster, default 1.2)
output_path: Output video path (auto-generated if None)
min_narration_words: Min narration length (generate mode only)
@@ -219,6 +223,8 @@ class VideoGeneratorService:
video_height=video_height,
video_fps=video_fps,
voice_id=voice_id,
tts_workflow=tts_workflow,
tts_speed=tts_speed,
image_width=image_width,
image_height=image_height,
image_workflow=image_workflow,
@@ -259,7 +265,8 @@ class VideoGeneratorService:
# Override prompt_prefix if provided (temporarily modify config)
original_prefix = None
if prompt_prefix is not None:
image_config = self.core.config.get("image", {})
# Fix: image config is under comfyui.image, not directly under config
image_config = self.core.config.get("comfyui", {}).get("image", {})
original_prefix = image_config.get("prompt_prefix")
image_config["prompt_prefix"] = prompt_prefix
logger.info(f"Using custom prompt_prefix: '{prompt_prefix}'")