Update app.py

app.py

@@ -11,9 +11,9 @@ from pydub import AudioSegment
 app = FastAPI()
 
 def split_text(text, max_chunk_size=500):
-    """Split text into chunks only if it's longer than max_chunk_size."""
+    """Split text into chunks if it exceeds max_chunk_size."""
     if len(text) <= max_chunk_size:
-        return [text]  # No need to split if it's within the limit
+        return [text]  
 
     sentences = text.replace('।', '.').replace('؟', '?').split('.')
     chunks = []
@@ -38,8 +38,8 @@ def split_text(text, max_chunk_size=500):
     return chunks
 
 async def process_chunk(text, voice, temp_dir, chunk_index):
-    """Process a single chunk of text into an MP3 file."""
-    tmp_path = os.path.join(temp_dir, f"chunk_{chunk_index}_{int(time.time())}_{os.urandom(4).hex()}.mp3")
+    """Process a single chunk into an MP3 file."""
+    tmp_path = os.path.join(temp_dir, f"chunk_{chunk_index}_{int(time.time())}.mp3")
     communicate = edge_tts.Communicate(text, voice)
     await communicate.save(tmp_path)
     return tmp_path
@@ -56,7 +56,7 @@ async def combine_audio_files(chunk_files):
     combined.export(output, format="mp3")
     output.seek(0)
 
-    # Clean up temp files
+    # Cleanup chunk files
     for file in chunk_files:
         try:
             os.remove(file)
@@ -74,26 +74,34 @@ async def tts(text: str, voice: str = "en-US-AriaNeural"):
     if not text.strip():
         return {"error": "Text cannot be empty."}
 
-    text_chunks = split_text(text)  # Split only if necessary
+    text_chunks = split_text(text)  # Only splits if text > 500 characters
 
-    if len(text_chunks) == 1:
-        # Process the entire text as a single request if it's within limit
-        output_audio = io.BytesIO()
-        communicate = edge_tts.Communicate(text_chunks[0], voice)
-        await communicate.save(output_audio)
-        output_audio.seek(0)
-
-        return StreamingResponse(output_audio, media_type="audio/mpeg", headers={"Content-Disposition": "attachment; filename=speech.mp3"})
-    
-    # If text is split into chunks, process them individually
-    with TemporaryDirectory() as temp_dir:
-        chunk_files = await asyncio.gather(*[
-            process_chunk(chunk, voice, temp_dir, i) for i, chunk in enumerate(text_chunks)
-        ])
+    async def event_stream():
+        """Send real-time status updates to the client while processing."""
+        yield "Processing started...\n"
         
-        output_audio = await combine_audio_files(chunk_files)
-
-    return StreamingResponse(output_audio, media_type="audio/mpeg", headers={"Content-Disposition": "attachment; filename=speech.mp3"})
+        if len(text_chunks) == 1:
+            # Single request processing
+            output_audio = io.BytesIO()
+            communicate = edge_tts.Communicate(text_chunks[0], voice)
+            await communicate.save(output_audio)
+            output_audio.seek(0)
+            yield "Processing completed. Downloading audio...\n"
+            yield output_audio.read()
+            return
+
+        with TemporaryDirectory() as temp_dir:
+            # Process all chunks concurrently
+            tasks = [process_chunk(chunk, voice, temp_dir, i) for i, chunk in enumerate(text_chunks)]
+            chunk_files = await asyncio.gather(*tasks)
+            
+            yield f"Processing {len(text_chunks)} chunks completed. Merging audio...\n"
+            output_audio = await combine_audio_files(chunk_files)
+
+        yield "Merging completed. Downloading final audio...\n"
+        yield output_audio.read()
+
+    return StreamingResponse(event_stream(), media_type="audio/mpeg")
 
 if __name__ == "__main__":
     import uvicorn