sherpa-onnx

sherpa-onnx copied to clipboard

refactor and pybind of OnlineWebsocketServer

6

Hi @csukuangfj , I have refactored the previous implementation, Below code base is functional tested and cleaned up I think you can suggest more for naming for more clarity or...

manickavela29

Keyword Spotter not detecting keywords - always returns null/empty

8

I need assistance with understanding why my mic cant detect any keywords : this is how i set it up , howvever everytime i try to print the keyword it...

dlangamandla

无法区分连字符和破折号，朗读错误

2

连字符（hyphen）​​ （Unicode: U+2010，键盘直接输入减号键 -） 如： short-term risks with long-term rewards. 连字符（hyphen -）用于连接单词或音节，形成复合词或拆分换行，朗读时需​​连贯无停顿 破折号（Em Dash — 或 ——）​​通常需要明显停顿 目前的情况是：在朗读英文文本时， 对连字符有明显停顿，对破折号没有停顿，这和正确读法刚好相反 （sherpa-onnx-non-streaming-tts-x64-v1.11.3.exe） 

WenBing2078

Dear sherpa-onnx Developers, I am a Flutter developer and I am very interested in integrating the powerful speech processing capabilities of sherpa-onnx, particularly Voice Activity Detection (VAD) and speaker recognition,...

hbbbs

請問 flutter-examples/streaming_asr 的 online_model.dart 如何設置 paraformer 模型？

1

 我想使用[sherpa-onnx-streaming-paraformer-bilingual-zh-en.tar.bz2](https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-streaming-paraformer-bilingual-zh-en.tar.bz2)的模型，但沒有範例中的joiner  請問該怎麼辦?

Chen566

TTS scaling pause not working?

5

After seeing this merged PR (https://github.com/k2-fsa/sherpa-onnx/pull/1820) I thought pauses could be controlled. But seems not likely. In that PR there is a new config parameter named "silence_scale", but it seems...

juangon

关于 如何在 android keyword-spotter 应用中 ，文本输入框设置自定义唤醒词？

7

我看在输入框有 Input your keywords here, one keyword per line.\nTwo example keywords are given below:\n\nn ǐ h ǎo @你好\nd àn g ē d àn g ē @蛋哥蛋哥 ，不是很理解 自定义关键字需要 添加 n...

XTaoWang

[BUG] OnlineRecognizer not outputting certain tokens after endpoint detection

3

I'm testing speech recognition from a microphone with endpoint detection using the provided Python [example](https://github.com/k2-fsa/sherpa-onnx/blob/master/python-api-examples/speech-recognition-from-microphone-with-endpoint-detection.py), and I've found the following issue: when the first token of a segment is the...

lgarcia-trebe

## 问题 目前的算法是如果识别到1.5秒静默帧或者识别到唤醒词才会reset清除之前识别出来的token，但是如果有一个唤醒词没有识别出来，接下来相似的唤醒词就有很大几率无法识别。 ## 建议 每次beam search仅搜索当前帧和之前一定时间内(如2秒)的结果

codepubs

如何使用build-ios后的产物

1

使用的部分应该有两个, 但是这个地方工程配置上看看需不需要补充一下  我看没有对应的文档，但是对应的ncnn的有一个文档 https://k2-fsa.github.io/sherpa/ncnn/ios/for-the-more-curious-swift.html 看看这部分是不是对onnx的部分可以补充一下

luckysmg