diff --git a/wechat_ipad/client/message.py b/wechat_ipad/client/message.py
index cb33fa0..405cb60 100644
--- a/wechat_ipad/client/message.py
+++ b/wechat_ipad/client/message.py
@@ -344,6 +344,7 @@ class MessageMixin(WechatAPIClientBase):
             self.logging.debug(f"2audio.frame_rate: {audio.frame_rate}")
             audio = audio.set_frame_rate(16000)
             self.logging.debug(f"3audio.frame_rate: {audio.frame_rate}")
+            audio = audio.set_channels(1).set_sample_width(2)  # 16-bit PCM
             logger.info(
                 f"音频处理: 格式={format}, 采样率={audio.frame_rate}, 声道数={audio.channels}, 时长={len(audio) / 1000}s")
             voice_base64 = base64.b64encode(