diff --git a/demo/components/audio_test.py b/demo/components/audio_test.py
index 542d9514c9fd94341cf2e7150cbd078a2dfa70c5..e56e36388ffaea0839c9b6de8867cf4c0591d999 100644
--- a/demo/components/audio_test.py
+++ b/demo/components/audio_test.py
@@ -14,35 +14,13 @@
 # =======================================================================
 # coding=utf-8
 from visualdl import LogWriter
-import numpy as np
-import wave
-
-
-def read_audio_data(audio_path):
-    """
-    Get audio data.
-    """
-    CHUNK = 4096
-    f = wave.open(audio_path, "rb")
-    rate = f.getframerate()
-    width = f.getsampwidth()
-    channel = f.getnchannels()
-    wavdata = []
-    chunk = f.readframes(CHUNK)
-
-    while chunk:
-        data = np.frombuffer(chunk, dtype='uint8')
-        wavdata.extend(data)
-        chunk = f.readframes(CHUNK)
-    shape = [rate, width, channel]
-    return shape, wavdata
+from scipy.io import wavfile
 
 
 if __name__ == '__main__':
-    with LogWriter(logdir="vdl_audio_0713") as writer:
-        audio_shape, audio_data = read_audio_data("./test.wav")
-        audio_data = np.array(audio_data)
+    with LogWriter(logdir="./log/audio_test/train") as writer:
+        sample_rate, audio_data = wavfile.read('./test.wav')
         writer.add_audio(tag="audio_tag",
                          audio_array=audio_data,
                          step=0,
-                         sample_rate=audio_shape[0])
+                         sample_rate=sample_rate)
diff --git a/demo/components/test.wav b/demo/components/test.wav
index a1170d8057d76ec1073a13089903e85332fe47ea..9c862b8df8af6ee0bf3387fb6f9a84ebe64735db 100644
Binary files a/demo/components/test.wav and b/demo/components/test.wav differ
diff --git a/visualdl/component/base_component.py b/visualdl/component/base_component.py
index ffc9c6c8300218c04675940b436cd2e0cd6b471b..1265aceac0e69ded3deeb785ad61ea74b31f5fab 100644
--- a/visualdl/component/base_component.py
+++ b/visualdl/component/base_component.py
@@ -131,15 +131,26 @@ def audio(tag, audio_array, sample_rate, step, walltime):
     Return:
         Package with format of record_pb2.Record
     """
+    audio_array = audio_array.squeeze()
+    if abs(audio_array).max() > 1:
+        print('warning: audio amplitude out of range, auto clipped.')
+        audio_array = audio_array.clip(-1, 1)
+    assert (audio_array.ndim == 1), 'input tensor should be 1 dimensional.'
+
+    audio_array = [int(32767.0 * x) for x in audio_array]
+
     import io
     import wave
+    import struct
 
     fio = io.BytesIO()
     wave_writer = wave.open(fio, 'wb')
     wave_writer.setnchannels(1)
     wave_writer.setsampwidth(2)
     wave_writer.setframerate(sample_rate)
-    wave_writer.writeframes(audio_array)
+    audio_enc = b''
+    audio_enc += struct.pack("<" + "h" * len(audio_array), *audio_array)
+    wave_writer.writeframes(audio_enc)
     wave_writer.close()
     audio_string = fio.getvalue()
     fio.close()