feat: add timestamp set in send_tts_audio for align words with audio

Author: YiminW

interface/ten_ai_base/tts2.py

@@ -160,7 +160,7 @@ async def _process_input_queue(self, ten_env: AsyncTenEnv) -> None:
 
 
     async def send_tts_audio_data(
-        self, audio_data: bytes
+        self, audio_data: bytes, timestamp: int
     ) -> None:
         """End sending audio out."""
         try:
@@ -195,9 +195,11 @@ async def send_tts_audio_data(
                     // (bytes_per_sample * number_of_channels)
                 )
                 f.alloc_buf(len(combined_data))
+                f.set_timestamp(timestamp)
                 buff = f.lock_buf()
                 buff[:] = combined_data
                 f.unlock_buf(buff)
+                self.ten_env.log_debug(f"send audio frame, timestamp: {timestamp}, length: {len(combined_data)}")
                 await self.ten_env.send_audio_frame(f)
         except Exception as e:
             self.ten_env.log_error(