Upload 2 files

main.py

@@ -9,7 +9,13 @@ from typing import Optional
 
 app = FastAPI(title="VocalizeAPI")
 
+# --- LINUX FFMPEG FIX ---
+# On some HF containers, we need to explicitly tell pydub where ffmpeg is
+if shutil.which("ffmpeg"):
+    AudioSegment.converter = shutil.which("ffmpeg")
+
 def get_audio_info(file_path):
+    """Returns size in MB and duration in seconds."""
     size_mb = os.path.getsize(file_path) / (1024 * 1024)
     audio = AudioSegment.from_file(file_path)
     duration_sec = len(audio) / 1000.0
@@ -21,19 +27,25 @@ async def audio_to_text(
     lang: Optional[str] = Form(None)
 ):
     temp_dir = tempfile.mkdtemp()
-    input_ext = Path(file.filename).suffix.lower() or ".ogg"
+    # Ensure filename exists to avoid suffix errors
+    filename = file.filename if file.filename else "input.ogg"
+    input_ext = Path(filename).suffix.lower() or ".ogg"
+    
     input_path = os.path.join(temp_dir, f"input{input_ext}")
     wav_path = os.path.join(temp_dir, "converted.wav")
     
     try:
+        # 1. Save uploaded file
         with open(input_path, "wb") as buffer:
             shutil.copyfileobj(file.file, buffer)
 
+        # 2. Convert to WAV (Standardized for Google)
         audio = AudioSegment.from_file(input_path)
         audio = audio.set_channels(1).set_frame_rate(16000)
         audio.export(wav_path, format="wav")
         target_path = wav_path
 
+        # 3. Validate
         size_mb, duration = get_audio_info(target_path)
         
         if duration > 60:
@@ -41,18 +53,19 @@ async def audio_to_text(
         elif size_mb > 10:
             return {"status": False, "error": "Error: File is too large (>10MB).", "result": ""}
 
+        # 4. Transcribe
         r = sr.Recognizer()
         with sr.AudioFile(target_path) as source:
             r.adjust_for_ambient_noise(source, duration=0.5) 
             audio_data = r.record(source)
             
             try:
-                # --- FIXED LOGIC HERE ---
-                if lang and lang.strip():
-                    # Use specific language if provided
-                    text = r.recognize_google(audio_data, language=lang)
+                # Use provided lang if it exists and isn't just whitespace
+                actual_lang = lang.strip() if (lang and lang.strip()) else None
+                
+                if actual_lang:
+                    text = r.recognize_google(audio_data, language=actual_lang)
                 else:
-                    # Use default (English/Auto) if lang is None or empty string
                     text = r.recognize_google(audio_data) 
                 
                 return {"status": True, "error": None, "result": text}
@@ -64,7 +77,11 @@ async def audio_to_text(
 
     except Exception as e:
         print(f"Server Log Error: {e}")
-        return {"status": False, "error": "System Error: Processing failed.", "result": ""}
+        return {"status": False, "error": f"System Error: {str(e)}", "result": ""}
     finally:
+        # 5. Cleanup
         shutil.rmtree(temp_dir, ignore_errors=True)
 
+if __name__ == "__main__":
+    import uvicorn
+    uvicorn.run(app, host="0.0.0.0", port=7860)