WIP

app.py

@@ -265,49 +265,46 @@ def transcribe_youtube(url, return_timestamps, generate_subs, chunk_length_s=15,
             logger.error(f"Invalid URL scheme: {url}")
             raise gr.Error("URL must start with http:// or https://")
         
-        # Get audio URL from Sieve
+        # Get audio file from Sieve
         logger.info("Starting YouTube audio download via Sieve API...")
-        audio_url = download_youtube_audio(url)
-        logger.info(f"Successfully obtained audio URL from Sieve: {audio_url}")
-        
-        # Validate audio URL scheme
-        if not audio_url.startswith(('http://', 'https://')):
-            logger.error(f"Invalid audio URL scheme from Sieve: {audio_url}")
-            raise gr.Error("Invalid audio URL scheme received from Sieve")
+        audio_file = download_youtube_audio(url)
+        logger.info(f"Successfully obtained audio file from Sieve: {audio_file}")
         
         # Send request to API
         logger.info("Sending transcription request to API...")
-        response = requests.post(
-            f"{API_URL}/transcribe/url",
-            json={
-                "url": audio_url,
-                "timestamp_level": "sentence" if return_timestamps else None,
-                "task": "transcribe",
-                "chunk_length_s": chunk_length_s,
-                "batch_size": batch_size
-            },
-            timeout=1800
-        )
-        response.raise_for_status()
-        result = response.json()
-        logger.info("Successfully received response from API")
-        
-        # Log metadata
-        metadata = result.get("metadata", {})
-        logger.info(f"Transcription metadata: {metadata}")
-        logger.info(f"Transcription completed in {metadata.get('timing', {}).get('total_time', 0):.2f} seconds")
-        
-        # Generate subtitles if requested
-        srt_file = None
-        if generate_subs and return_timestamps and "segments" in result["transcription"]:
-            logger.info("Generating SRT subtitles...")
-            srt_content = generate_srt(result["transcription"]["segments"])
-            srt_file = save_srt_to_file(srt_content)
-            logger.info(f"Generated SRT file: {srt_file}")
-        
-        logger.info("YouTube transcription process completed successfully")
-        return result, srt_file, ""
-        
+        with open(audio_file, "rb") as f:
+            files = {"file": f}
+            response = requests.post(
+                f"{API_URL}/transcribe",
+                files=files,
+                data={
+                    "timestamp_level": "sentence" if return_timestamps else None,
+                    "task": "transcribe",
+                    "chunk_length_s": chunk_length_s,
+                    "batch_size": batch_size
+                },
+                timeout=1800
+            )
+            response.raise_for_status()
+            result = response.json()
+            logger.info("Successfully received response from API")
+            
+            # Log metadata
+            metadata = result.get("metadata", {})
+            logger.info(f"Transcription metadata: {metadata}")
+            logger.info(f"Transcription completed in {metadata.get('timing', {}).get('total_time', 0):.2f} seconds")
+            
+            # Generate subtitles if requested
+            srt_file = None
+            if generate_subs and return_timestamps and "segments" in result["transcription"]:
+                logger.info("Generating SRT subtitles...")
+                srt_content = generate_srt(result["transcription"]["segments"])
+                srt_file = save_srt_to_file(srt_content)
+                logger.info(f"Generated SRT file: {srt_file}")
+            
+            logger.info("YouTube transcription process completed successfully")
+            return result, srt_file, ""
+            
     except Exception as e:
         logger.exception(f"Error in YouTube transcription: {str(e)}")
         raise gr.Error(f"Failed to transcribe YouTube video: {str(e)}")