From 549c2af9884272365f7005b00757c3a7aa8ae551 Mon Sep 17 00:00:00 2001 From: Jad Date: Sun, 30 Mar 2025 13:06:02 +0800 Subject: [PATCH] =?UTF-8?q?feat(server):=20asr=E4=BF=9D=E7=95=99=E6=9B=B4?= =?UTF-8?q?=E5=A4=9A=E7=9A=84=E9=9F=B3=E9=A2=91=E5=89=8D=E5=AF=BC=E5=B8=A7?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit 语速较快或者声音不清晰时，保留更多前导帧可以提升asr识别效果 --- main/xiaozhi-server/core/handle/receiveAudioHandle.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/main/xiaozhi-server/core/handle/receiveAudioHandle.py b/main/xiaozhi-server/core/handle/receiveAudioHandle.py index c5b4135c..f97e1e23 100644 --- a/main/xiaozhi-server/core/handle/receiveAudioHandle.py +++ b/main/xiaozhi-server/core/handle/receiveAudioHandle.py @@ -21,7 +21,7 @@ async def handleAudioMessage(conn, audio): if have_voice == False and conn.client_have_voice == False: await no_voice_close_connect(conn) conn.asr_audio.append(audio) - conn.asr_audio = conn.asr_audio[-5:] # 保留最新的5帧音频内容，解决ASR句首丢字问题 + conn.asr_audio = conn.asr_audio[-10:] # 保留最新的10帧音频内容，解决ASR句首丢字问题 return conn.client_no_voice_last_time = 0.0 conn.asr_audio.append(audio) @@ -30,7 +30,7 @@ async def handleAudioMessage(conn, audio): conn.client_abort = False conn.asr_server_receive = False # 音频太短了，无法识别 - if len(conn.asr_audio) < 10: + if len(conn.asr_audio) < 15: conn.asr_server_receive = True else: text, file_path = await conn.asr.speech_to_text(conn.asr_audio, conn.session_id)