Offline speech to text (Japanese voice)

[cuongttn2]

@csukuangfj
Hi, I downloaded this sherpa-onnx version: (sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17),
I followed the instructions when unzipping the folder, it seems the file names are not consistent with the instructions, Can you or someone guide me how to install it properly, Sorry because I am really bad at approaching SenseVoice
Thanks a lot

[csukuangfj]

please describe in detail the issue you have.

[csukuangfj]

if you want to use Android with this model, please use
https://github.com/k2-fsa/sherpa-onnx/tree/master/android%2FSherpaOnnxVadAsr

I also recommend the ReazonSpeech trained model in the doc for Japanese.

[cuongttn2]

I am trying to create an Android app demo with the main feature of converting voice (Japanese) to text (Japanese text or characters) via microphone and no internet required (offline).
I followed the instructions with this sherpa model (sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17. There is a difficulty for me that the files in the tutorial are not in the sherpa model that I downloaded, can I do it with any tutorial or documentation for Android demo? Hope to get help

please describe in detail the issue you have.

[csukuangfj]

please see the link I posted before.

I suggest that you read the beginning of the doc carefully and know how to adapt it to your own need.

I hope.you know.that the Android doc.is for real time speech recognition. But the method described in the doc applies equally well to non-streaming speech recognition.

I also hope you.know the difference between streaming and non-streaming ASR. SenseVoice is a non-streaming model.

Please don't follow the Android doc blindly. You MUST adapt it by yourself for.VadAsr.