Update app.py

app.py

@@ -32,6 +32,10 @@ class TranslationJob(BaseModel):
     supabase_anon_key: Optional[str] = None
     gemini_api_key: Optional[str] = None
     elevenlabs_api_key: Optional[str] = None
+    use_voice_cloning: Optional[bool] = True
+    default_voice_id: Optional[str] = 'Xb7hH8MSUJpSbSDYk0k2'
+    trim_start: Optional[float] = None
+    trim_end: Optional[float] = None
 
 def load_env_local():
     """Load environment variables from local .env.local file if it exists."""
@@ -374,11 +378,18 @@ def run_production_pipeline(job: TranslationJob, gemini_key: str, elevenlabs_key
         # 1.5. Standardize video (H.264/AAC, 25 FPS, YUV420p) for OpenCV / Wav2Lip stability
         print(f"[{video_id}] Standardizing downloaded video to MP4 format...")
         cmd_standardize = [
-            "ffmpeg", "-y", "-i", downloaded_video_path,
+            "ffmpeg", "-y", "-i", downloaded_video_path
+        ]
+        if job.trim_start is not None and job.trim_start > 0:
+            cmd_standardize.extend(["-ss", str(job.trim_start)])
+        if job.trim_end is not None and job.trim_end > 0:
+            cmd_standardize.extend(["-to", str(job.trim_end)])
+            
+        cmd_standardize.extend([
             "-c:v", "libx264", "-pix_fmt", "yuv420p", "-r", "25",
             "-c:a", "aac", "-ar", "16000", "-ac", "1",
             input_video_path
-        ]
+        ])
         subprocess.run(cmd_standardize, check=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
         
         # 2. Extract audio from video using FFmpeg
@@ -438,11 +449,53 @@ def run_production_pipeline(job: TranslationJob, gemini_key: str, elevenlabs_key
         
         # 4. Synthesize voice cloning using ElevenLabs API
         print(f"[{video_id}] Calling ElevenLabs TTS Voice Cloning with tone '{job.voice_tone}'...")
-        # We can use the default standard voice (e.g. Rachel: 21m00Tcm4TlvDq8ikWAM)
-        voice_id = "21m00Tcm4TlvDq8ikWAM" 
-        elevenlabs_url = f"https://api.elevenlabs.io/v1/text-to-speech/{voice_id}"
         
         elevenlabs_headers = {
+            "xi-api-key": elevenlabs_key
+        }
+        
+        voice_id = None
+        cloned_voice_created = False
+        
+        # Try to dynamically clone the voice using the extracted audio if requested
+        if job.use_voice_cloning:
+            try:
+                print(f"[{video_id}] Attempting to create a temporary voice clone from extracted audio...")
+                add_voice_url = "https://api.elevenlabs.io/v1/voices/add"
+                
+                # Open audio file for upload
+                with open(extracted_audio_path, "rb") as audio_file:
+                    files = {
+                        "files": (os.path.basename(extracted_audio_path), audio_file, "audio/mpeg")
+                    }
+                    data = {
+                        "name": f"VoiceClips_{video_id}",
+                        "description": f"Temporary cloned voice for job {video_id}"
+                    }
+                    
+                    add_res = requests.post(add_voice_url, headers=elevenlabs_headers, files=files, data=data)
+                    
+                if add_res.status_code == 200:
+                    voice_id = add_res.json().get("voice_id")
+                    cloned_voice_created = True
+                    print(f"[{video_id}] Voice clone created successfully. Voice ID: {voice_id}")
+                else:
+                    print(f"[{video_id}] Voice cloning failed with status code {add_res.status_code}: {add_res.text}")
+                    print(f"[{video_id}] Falling back to standard pre-made voice ({job.default_voice_id}).")
+            except Exception as clone_err:
+                print(f"[{video_id}] Exception during voice cloning: {clone_err}")
+                print(f"[{video_id}] Falling back to standard pre-made voice ({job.default_voice_id}).")
+        else:
+            print(f"[{video_id}] Voice cloning disabled by user. Using standard pre-made voice ({job.default_voice_id}).")
+            
+        # Fallback to default voice if cloning wasn't successful
+        if not voice_id:
+            voice_id = job.default_voice_id or "Xb7hH8MSUJpSbSDYk0k2"  # Alice (pre-made voice, works on free tier)
+            
+        elevenlabs_url = f"https://api.elevenlabs.io/v1/text-to-speech/{voice_id}"
+        
+        # Content-Type header is needed for the JSON payload of the TTS request
+        tts_headers = {
             "xi-api-key": elevenlabs_key,
             "Content-Type": "application/json"
         }
@@ -466,12 +519,25 @@ def run_production_pipeline(job: TranslationJob, gemini_key: str, elevenlabs_key
             }
         }
         
-        tts_res = requests.post(elevenlabs_url, headers=elevenlabs_headers, json=elevenlabs_payload)
+        tts_res = requests.post(elevenlabs_url, headers=tts_headers, json=elevenlabs_payload)
         tts_res.raise_for_status()
         
         with open(synthesized_audio_path, "wb") as f:
             f.write(tts_res.content)
             
+        # If we successfully created a temporary cloned voice, delete it now to free up slots
+        if cloned_voice_created and voice_id:
+            try:
+                print(f"[{video_id}] Deleting temporary cloned voice {voice_id}...")
+                delete_url = f"https://api.elevenlabs.io/v1/voices/{voice_id}"
+                del_res = requests.delete(delete_url, headers=elevenlabs_headers)
+                if del_res.status_code == 200:
+                    print(f"[{video_id}] Temporary cloned voice deleted successfully.")
+                else:
+                    print(f"[{video_id}] Failed to delete voice: {del_res.text}")
+            except Exception as del_err:
+                print(f"[{video_id}] Exception deleting voice: {del_err}")
+            
         # 5. Merge synthesized audio back with the original video (or run Lip-Sync if selected)
         lipsync_output_path = f"temp_{video_id}_lipsync.mp4"