Fabric Microphone Text Input Mod User Documention

icon

Introduction

Functionality

This mod allows you to use your microphone to input text and send it as a player message. It is a client-side mod, so it does not require any server-side setup.

Operating Principles

This mod uses VOSK Offline Speech Recognition API to recognize the text from the microphone. The recognized text will be sent to the server as a player message when the user presses the V key.

What's New in 1.1.x

Use VOSK Offline Speech Recognition API to recognize text from the microphone instead of CMU Sphnix, so that the recognition is more accurate and the recognition speed is faster.
Add a configuration UI by importing MidnightLib so that you can configure the mod to choose the language model, and since 1.1.0 we will not provide the language model in the mod package, you need to download the language model from VOSK Models.
For more information, please visit Jaffe2718/Fabric-Microphone-Text-Input|README

User Guide

For 1.1.0 or later version

1. Installation

Download the appropriate version of Minecraft with Fabric loader;

Install the appropriate version of the Fabric API and MidnightLib and Mod Menu as dependencies;

Download the 1.1.x version of the mod from GitHub Releases or Modrinth;

Put the downloaded mod JAR file into the mods folder of the Minecraft installation directory;

2. Configuration

notice: Since 1.1.0, we will not provide the language model in the mod package, you need to download the language model from VOSK Models.This mod will not work properly without the correct configuration.

Download the language model from VOSK Models and extract it. The recommended path is ...\.minecraft\versions\YOUR_GAME_VERSION\.mcmti\models, if the folders do not exist, you need to create them.

Launch Minecraft and enter the game;

Click the Mods button in the main menu;

Click the settings button of the Fabric Microphone Text Input mod, Then you will enter the configuration UI of Fabric Microphone Text Input Mod;

Fill the absolute path of the language model like this:

and ensure to click the Done button to save the configuration.

Restart Minecraft and enter the game.

Enter the game and press V to recognize the text and automatically send as a chat message.

If the message is messy code, please try the error coding repair function, this is an example:

Warning: This function is an experimental function, and it may cause the recognition to fail.

For 1.0.x

Installation

Install Minecraft with Fabric Loader(>=0.14.9) and Fabric API;
Download the mod jar with appropriate version from GitHub Releases or Modrinth;
Copy the mod jar file to the mods folder in your Minecraft installation directory;
Run Minecraft and enjoy the mod.

How to Use

Press the v key to start listening to your voice;
Say something and the mod will convert your voice into text and send it to the game as a chat message;
The message will be automatically sent.

Tips:

The accuracy of speech recognition is related to the model called by the program, not to the program ontology. The more accurate the model, the more accurate the recognition. The model used by this mod is the default model provided by CMU Sphnix, which is not very accurate. If you want to use a more accurate model, you can download the model you want to use from CMU Sphnix or train your own model. Then you can change the model used by the mod by following the instructions in the Developer Guide. If you find a better acoustic language model, you can let me know at my GitHub Issues.

FAQ

Q: Why the program cannot to recognize the text from the microphone?

A: The program cannot recognize the text from the microphone may be caused by the following reasons:

The microphone cannot be used due to incorrect microphone settings on the computer;
Java cannot access the microphone due to incorrect Java permission settings;
The language model is not selected correctly;

Please confirm that the microphone properties of the computer are set correctly, that Java has permission to access the microphone, and that the absolute path configuration of the acoustic model is correct.

Q: Why speech recognition can recognize normally, but the output is messy code?

A: The output is messy code may refer to the default encoding of your operating system.

Please set the default encoding for your operating system to UTF-8. For 1.1.x, you can try the error coding repair function in this mod, but this function is an experimental function, and it may cause the recognition to fail, so we suggest you use this method with caution.

About

Repository

GitHub

License

This mod is licensed under the MIT License

Author

Jaffe2718

Issue

If you have any questions or suggestions, please submit an issue on GitHub Issues.

Contact

You can also contact me through the following methods: Bilibili Jaffe-

Microphone Text Input