sherpa-onnx

sherpa-onnx copied to clipboard

Android and iOS same model to extract the voice print feature vector comparison is inconsistent, the degree of acquaintance is very low

1

Big guys, Android and iOS have the same model to extract the same voiceprint feature vector comparison is inconsistent, the degree of acquaintance is very low, what may be caused...

xiexuetao

Android/flutter 调用sherpa-onnx包中的initBindings()方法导致应用闪退

6

背景：flutter构建Android应用 现象：应用直接闪退，无法正常运行 触发条件：调用initBindings()方法 错误信息： F/libc ( 9956): Fatal signal 11 (SIGSEGV), code 2 (SEGV_ACCERR), fault addr 0x710440b84618 in tid 9986 (1.ui), pid 9956 (sherpa.onnx.tts) （详情见图）  备注：把示例项目flutter-examples/tts下载下来debug构建运行也会出现一样的错误，基本上只要调用initBindings()方法的地方都会触发，其中一处参见下图 

SummerMonsoon1061

语音唤醒循环测试唤醒成功率偏低

4

sherpa-onnx已更新到最新版本1.10.44 使用一条音频循环播放500次，每次间隔 5 秒，测试环境为偏安静。使用keyword-spotter-from-microphone.py运行，唤醒率只有90%左右，理论上应该是100%的。同样，我基于keyword-spotter改写成伪流式的版本，唤醒率则达到100%。请问有什么潜在的原因导致实时唤醒的成功率下降吗？

TungyuYoung

I want to use the the following model `3dspeaker_speech_campplus_sv_en_voxceleb_16k.onnx ` to extract voice embeddings in a batch process. I don't think the `sherpa_onnx` python library allows me to do so,...

PhilipAmadasun

求大神指导一下unity编译安卓手机apk如何调用sheerpa-onnx

10

我是新手小白，想自己练练手，所以想用unity直接编译个手机app看看，我按照一位前辈的issue里面的操作步骤做了，但是刚开始提示找不到sherpa-onnx，我放了sherpa-onnx.dll到plugin下面，其他的都和前辈发文截图里面的差不多一样，然后在文件拷贝完成后就闪退，请大神指导一下unity编译安卓手机apk如何调用sheerpa-onnx，感谢！ Good Morning, i will be sharing the code needed to make sherpa-onnx tts work on unity specifically on android builds! the problem is that unity merges the model files...

newkitty

Restricted vocabulary for transducer models

4

For CTC-based models, we can use HLG.fst to restrict the ASR output to a predefined vocabulary, ensuring that only specific words are recognized. However, streaming CTC models do not generate...

rohithkodali

[WIP] add CTC prefix beam search / hotwords / shallow fussion

2

This PR implements the core part (c++/python/JNI) of CTC prefix beam search related decoding methods, including hotwords and rnnlm shallow fussion. - [x] offline prefix beam search - [x] offline...

pkufool

vad fails with cuda provider

3

using example vad with cpu provider works when using cuda provider, if tails with following, tested on docker * nvidia/cuda:11.8.0-cudnn8-devel-ubuntu22.04 * 1.10.45+cuda ```bash asr-firered-1 | 2025-02-20 02:51:23.338585289 [E:onnxruntime:, cuda_call.cc:116 CudaCall]...

DeoLeung

能否添加支持timestamp模型

3

如speech_timestamp_prediction-v1-16k-offline等模型专用于时间预测。。

woowloop

example about updating vad config in python

1

hi, is the config update function exposed in python api for vad models(silero) and how to use it? it seems not as i see in the source code :(

DeoLeung