Skip to content

README_en.md

Latest commit

 

History

History
117 lines (85 loc) · 3.75 KB

README_en.md

File metadata and controls

117 lines (85 loc) · 3.75 KB

TTSClient

[日本語] [English] [한국어] [中文简体]

This is client software for real-time speech recognition/Speech To Text/Automatic Speech Recognition (STT/ASR).

Recognized text can be obtained in real-time via WebHook or WebSocket.

We plan to support various AI services.

What's New

  • v.1.0.4 released
    • source code is opened.

Related Software

Download

Hugging Face repository.

  • win_std Edition: This is the edition for Windows that runs on a CPU. It is slower compared to the CUDA version, but should work on recent CPUs with decent specs.
  • win_cuda Edition: This is the edition for Windows that runs on an NVIDIA GPU. It operates faster due to GPU hardware acceleration.
  • mac Edition: This is the edition for Mac(Apple silicon(M1, M2, M3, etc)).

Online Demo

You can run this on Colab. Please access this notebook and click the Open in Colab button in the top left to open it.

Usage

  • After extracting the zip file, execute start_http.bat. Then, access the displayed URL in a browser.
  • Use start_https.bat to access the software remotely.
  • (For advanced users) Use start_http_with_ngrok.bat to access the software via ngrok tunneling.

Note: For the mac edition, replace .bat with .command.

How to Clone and Use the Repository (Advanced)

Prerequisites

Preparation

Please execute the following commands:

git clone https://github.com/w-okada/asrclient.git

cd asrclient
cd third_party
git clone https://github.com/reazon-research/ReazonSpeech
cd ..

poetry install

Execution

Please execute the following command:

poetry run main cui

Please access the URL in the image below. image

If you encounter an error like fatal error: Python.h: No such file or directory, please install the Python source code.

sudo apt install python3-dev

If you encounter an error like No such file or directory: 'cmake', install cmake with:

sudo apt-get install cmake

If using NVIDIA GPU (CUDA)

poetry remove onnxruntime-directml torch-directml
poetry add torch==2.3.1 --source torch_cuda12
poetry add onnxruntime-gpu==1.17.1 --source ort_cuda12

Webhook Client Sample

Python

Please execute the following command and set the displayed webhook URL in the GUI.

python .\client\webhook_server\generic_main.py

Node

Please execute the following command and set the displayed webhook URL in the GUI.

node .\client\webhook_server\generic_main.js

Socket.io Client Sample

Please execute the following command. Replace http://localhost:20000 with the URL where ASRClient is running.

poetry run python_socket_io http://localhost:20000
⚡