up33

app.py

@@ -4,6 +4,7 @@ import outetts
 import io
 import json
 import base64
+import struct
 import os 
 # Initialize the interface
 interface = outetts.Interface(
@@ -54,8 +55,32 @@ async def websocket_tts(websocket: WebSocket):
             chunk_size = 4096
             try:
                 with open(temp_path, "rb") as f:
-                    while True:
-                        chunk = f.read(chunk_size)
+                    wav_data = f.read()
+                    # WAV header is typically 44 bytes, but let's detect it robustly
+                    # Find the end of the header (data chunk)
+                    if wav_data[:4] != b'RIFF' or wav_data[8:12] != b'WAVE':
+                        raise ValueError("Not a valid WAV file")
+                    # Find 'data' subchunk
+                    data_offset = wav_data.find(b'data')
+                    if data_offset == -1:
+                        raise ValueError("No 'data' chunk found in WAV file")
+                    header_end = data_offset + 8  # 'data' + size (4 bytes)
+                    wav_header = wav_data[:header_end]
+                    pcm_data = wav_data[header_end:]
+                    # Send header + first PCM chunk
+                    first_chunk = pcm_data[:chunk_size]
+                    audio_b64 = base64.b64encode(wav_header + first_chunk).decode("ascii")
+                    await websocket.send_text(json.dumps({
+                        "data": {
+                            "audio_bytes": audio_b64,
+                            "duration": None,
+                            "request_finished": False
+                        }
+                    }))
+                    # Send rest of PCM data in chunks (without header)
+                    idx = chunk_size
+                    while idx < len(pcm_data):
+                        chunk = pcm_data[idx:idx+chunk_size]
                         if not chunk:
                             break
                         audio_b64 = base64.b64encode(chunk).decode("ascii")
@@ -66,6 +91,7 @@ async def websocket_tts(websocket: WebSocket):
                                 "request_finished": False
                             }
                         }))
+                        idx += chunk_size
             finally:
                 try:
                     os.remove(temp_path)