3

fixed_app.py

@@ -32,6 +32,7 @@ class AudioRequest(BaseModel):
     beam_size: Optional[int] = 1
     condition_on_previous_text: Optional[bool] = False
     fast_mode: Optional[bool] = True  # 快速模式
+    vad: Optional[bool] = False
 
 def load_model(model_name: str):
     """确保模型文件存在，返回模型路径"""
@@ -245,8 +246,8 @@ async def transcribe_audio(request: AudioRequest):
                 "-ac", "0",  # 音频上下文 = 0 (最快)
                 "-nf",       # --no-fallback: 禁用温度回退
                 "-tp", "0.0", # 温度 = 0 (确定性输出)
-                "--vad",
-                "-vm", "/app/models/ggml-silero-v5.1.2.bin"
+                "--vad" if request.vad else "",
+                "-vm", "/app/models/ggml-silero-v5.1.2.bin" if request.vad else ""
             ]
         else:
             # 标准模式：平衡速度和精度
@@ -311,6 +312,7 @@ async def transcribe_audio(request: AudioRequest):
             
             # 读取JSON输出文件
             result = parse_whisper_output(output_file, stdout, exit_code)
+            result["processing_time"] = processing_time
             
             return result
 

startup.sh

@@ -1,7 +1,7 @@
 #!/bin/bash
 
 # 显示环境信息
-echo "=== Whisper API Startup 0.2==="
+echo "=== Whisper API Startup 0.3==="
 echo "Python version: $(python3 --version)"
 echo "Current directory: $(pwd)"
 # echo "Files in /app:"