Page 1 of 1
ASR例程，语音识别准确度不高
Posted: Tue Aug 02, 2022 6:45 am
by wangpengXX
离线语音识别不太准确，有时要说好几次才识别，有时说了没反应。感觉不灵敏啊
Re: ASR例程，语音识别准确度不高
Posted: Wed Aug 03, 2022 3:07 am
by ESP_HengYC
选择 wakenet 唤醒等级高的配置试试， 例如：
Code: Select all
[*] use wakenet
Wake word engine (WakeNet 5 (quantized with 16-bit)) --->
Wake word name (Hi,Lexin (WakeNet5X3)) --->
[*] use multinet
langugae (chinese) --->
speech commands recognition model (chinese single recognition (MultiNet2)) --->
Add speech commands --->
Re: ASR例程，语音识别准确度不高
Posted: Thu Aug 04, 2022 6:11 am
by wangpengXX
唤醒每次都可以，是识别不灵敏，提高wakenet有用吗
Re: ASR例程，语音识别准确度不高
Posted: Fri Jan 06, 2023 2:21 am
by dodo55
wangpengXX wrote: ↑Tue Aug 02, 2022 6:45 am
离线语音识别不太准确，有时要说好几次才识别，有时说了没反应。感觉不灵敏啊
请问你用的是具体哪个例程，用啥开发板？
Re: ASR例程，语音识别准确度不高
Posted: Mon Feb 06, 2023 8:56 am
by ESP_William
关于识别
建议使用最新版本的 ESP- ADF+ESP32-S3-Korvo-2 来测试，以及配置选择高版本的 hilexin 语音库
效果会比较好