diff --git a/python/paddle/audio/backends/backend.py b/python/paddle/audio/backends/backend.py
index 87fd217ee37ab5..13afa8fd786d4e 100644
--- a/python/paddle/audio/backends/backend.py
+++ b/python/paddle/audio/backends/backend.py
@@ -12,15 +12,25 @@
 # See the License for the specific language governing permissions and
 # limitations under the License
 
-from pathlib import Path
-from typing import Optional, Tuple, Union
+from __future__ import annotations
 
-import paddle
+from typing import TYPE_CHECKING, BinaryIO
+
+if TYPE_CHECKING:
+    from pathlib import Path
+
+    from paddle import Tensor
 
 
 class AudioInfo:
     """Audio info, return type of backend info function"""
 
+    sample_rate: int
+    num_samples: int
+    num_channels: int
+    bits_per_sample: int
+    encoding: str
+
     def __init__(
         self,
         sample_rate: int,
@@ -28,7 +38,7 @@ def __init__(
         num_channels: int,
         bits_per_sample: int,
         encoding: str,
-    ):
+    ) -> None:
         self.sample_rate = sample_rate
         self.num_samples = num_samples
         self.num_channels = num_channels
@@ -36,11 +46,11 @@ def __init__(
         self.encoding = encoding
 
 
-def info(filepath: str) -> AudioInfo:
+def info(filepath: str | BinaryIO) -> AudioInfo:
     """Get signal information of input audio file.
 
     Args:
-       filepath: audio path or file object.
+        filepath: audio path or file object.
 
     Returns:
         AudioInfo: info of the given audio.
@@ -68,12 +78,12 @@ def info(filepath: str) -> AudioInfo:
 
 
 def load(
-    filepath: Union[str, Path],
+    filepath: str | Path,
     frame_offset: int = 0,
     num_frames: int = -1,
     normalize: bool = True,
     channels_first: bool = True,
-) -> Tuple[paddle.Tensor, int]:
+) -> tuple[Tensor, int]:
     """Load audio data from file.Load the audio content start form frame_offset, and get num_frames.
 
     Args:
@@ -113,12 +123,12 @@ def load(
 
 def save(
     filepath: str,
-    src: paddle.Tensor,
+    src: Tensor,
     sample_rate: int,
     channels_first: bool = True,
-    encoding: Optional[str] = None,
-    bits_per_sample: Optional[int] = 16,
-):
+    encoding: str | None = None,
+    bits_per_sample: int | None = 16,
+) -> None:
     """
     Save audio tensor to file.
 
diff --git a/python/paddle/audio/backends/init_backend.py b/python/paddle/audio/backends/init_backend.py
index ada7e21dc3e60c..d3bad4a283a1f6 100644
--- a/python/paddle/audio/backends/init_backend.py
+++ b/python/paddle/audio/backends/init_backend.py
@@ -12,9 +12,10 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+from __future__ import annotations
+
 import sys
 import warnings
-from typing import List
 
 import paddle
 
@@ -34,11 +35,11 @@ def _check_version(version: str) -> bool:
     return True
 
 
-def list_available_backends() -> List[str]:
+def list_available_backends() -> list[str]:
     """List available backends, the backends in paddleaudio and the default backend.
 
     Returns:
-        List[str]: The list of available backends.
+        list[str]: The list of available backends.
 
     Examples:
         .. code-block:: python
@@ -136,7 +137,7 @@ def get_current_backend() -> str:
     return "wave_backend"
 
 
-def set_backend(backend_name: str):
+def set_backend(backend_name: str) -> None:
     """Set the backend by one of the list_audio_backend return.
 
     Args:
@@ -188,7 +189,7 @@ def set_backend(backend_name: str):
         setattr(paddle.audio, func, getattr(module, func))
 
 
-def _init_set_audio_backend():
+def _init_set_audio_backend() -> None:
     # init the default wave_backend.
     for func in ["save", "load", "info"]:
         setattr(backend, func, getattr(wave_backend, func))
diff --git a/python/paddle/audio/backends/wave_backend.py b/python/paddle/audio/backends/wave_backend.py
index 08e107ed3c8dcf..4b3c8f3eaee787 100644
--- a/python/paddle/audio/backends/wave_backend.py
+++ b/python/paddle/audio/backends/wave_backend.py
@@ -12,9 +12,10 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+from __future__ import annotations
+
 import wave
-from pathlib import Path
-from typing import Optional, Tuple, Union
+from typing import TYPE_CHECKING, BinaryIO
 
 import numpy as np
 
@@ -22,6 +23,11 @@
 
 from .backend import AudioInfo
 
+if TYPE_CHECKING:
+    from pathlib import Path
+
+    from paddle import Tensor
+
 
 def _error_message():
     package = "paddleaudio"
@@ -34,11 +40,11 @@ def _error_message():
     return warn_msg
 
 
-def info(filepath: str) -> AudioInfo:
+def info(filepath: str | BinaryIO) -> AudioInfo:
     """Get signal information of input audio file.
 
     Args:
-       filepath: audio path or file object.
+        filepath: audio path or file object.
 
     Returns:
         AudioInfo: info of the given audio.
@@ -87,12 +93,12 @@ def info(filepath: str) -> AudioInfo:
 
 
 def load(
-    filepath: Union[str, Path],
+    filepath: str | Path,
     frame_offset: int = 0,
     num_frames: int = -1,
     normalize: bool = True,
     channels_first: bool = True,
-) -> Tuple[paddle.Tensor, int]:
+) -> tuple[Tensor, int]:
     """Load audio data from file. load the audio content start form frame_offset, and get num_frames.
 
     Args:
@@ -167,12 +173,12 @@ def load(
 
 def save(
     filepath: str,
-    src: paddle.Tensor,
+    src: Tensor,
     sample_rate: int,
     channels_first: bool = True,
-    encoding: Optional[str] = None,
-    bits_per_sample: Optional[int] = 16,
-):
+    encoding: str | None = None,
+    bits_per_sample: int | None = 16,
+) -> None:
     """
     Save audio tensor to file.