Skip to content

A speech-to-text framework and bot for Discord. Take control of your Discord server using speech and voice commands. Can also be useful for hearing impaired and deaf people.

License

Notifications You must be signed in to change notification settings

inevolin/DiscordEarsBot

Repository files navigation

DiscordEarsBot

A speech-to-text bot for Discord written in NodeJS. Can be useful for hearing impaired and deaf people.

Getting Started:

Discord Ears Bot Demo

Installation Tutorial

YouTube: https://www.youtube.com/watch?v=IKIlnaCDZcI

Try the bot for yourself on our Discord server: https://discord.gg/ApdTMG9

Developers

Setup

In your Discord Developers Bot settings, you need to enable these intents:

image

Manual Installation

You need nodeJS version +16 with npm on your machine, use node -v to check your version. Execute the following commands:

git clone https://github.com/healzer/DiscordEarsBot.git
cd DiscordEarsBot
npm install

Provide the Discord API Token using DISCORD_TOK Env Variable or in settings.json.

Finally run node index.js. You can also use pm2 or nodemon to keep the bot running 24/7.

Heroku

If you don't have a linux server/machine then you can use Heroku for hosting your bot 24/7 and it's free.

  1. Fork this GitHub repository
  2. Create Discord Bot, Invite it to your server and get the API Token
  3. Create new Heroku app, use the GitHub method and Deploy DiscordEarsBot
  4. Under "resources" disable "web" and enable "worker" dyno instead.
  5. Provide the DISCORD_TOK Config Var under "settings"

Usage

By now you have a discord server, the DiscordEarsBot is running and is a part of your server. Make sure your server has a text and voice channel.

  1. Enter one of your voice channels.
  2. In one of your text channels type: *join, the bot will join the voice channel.
  3. Everything said within that channel will be transcribed into text (as long as the bot is within the voice channel).
  4. Type *leave to make the bot leave the voice channel.
  5. Type *help for a list of commands.

notes:

  • When the bot is inside a voice channel it listens to all speech and transcribes audio into text.
  • Each user is a separate audio channel, the bot hears everyone separately.
  • Only when your user picture turns green in the voice channel will the bot receive your audio.
  • A long pause interrupts the audio input.
  • For Google Speech & WitAI: The duration of a single audio input is limited to 20 seconds, longer audio is not transcribed.

Speech Services

Speech to Text Discord Bot

YouTube comparison and tutorial for developers on choosing the right Speech API: https://www.youtube.com/watch?v=fQcEZIgw_LA

Vosk API

This is our default Speech-to-Text method. The Vosk API is a free & open-source solution that runs locally (offline). By default only english is enabled. Developers can change or include more language models from here: https://alphacephei.com/vosk/models

WitAI

Installation:

  1. set SPEECH_METHOD to witai
  2. use your Server Access Token for WITAI_TOK

WitAI supports over 120 languages (https://wit.ai/faq), however only one language can be used at a time. If you're not speaking English on Discord, then change your default language on WitAI under "settings" for your app.

You can also change the language using the following bot command: *lang <code> <code> should be an ISO 639-1 language code (2 digits): https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes

Google Speech API

You can use Google's Speech-to-Text API as follows:

  1. set SPEECH_METHOD to google
  2. For non-English transcriptions: open index.js, inside the function transcribe_gspeech change the value of languageCode.
  3. Enable Google Speech API here: https://console.cloud.google.com/apis/library/speech.googleapis.com
  4. Create a new Service Account (or use your existing one): https://console.cloud.google.com/apis/credentials
  5. Create a new Service Account Key (or use existing) and download the json file.
  6. Put the json file inside your bot directory and rename it to gspeech_key.json.

Mozilla DeepSpeech (experimental)

Using Mozilla DeepSpeech for speech recognition, tutorial.

Contact

For enquiries or issues get in touch with me:

Name: Ilya Nevolin

Email: [email protected]

Discord: https://discord.gg/ApdTMG9

About

A speech-to-text framework and bot for Discord. Take control of your Discord server using speech and voice commands. Can also be useful for hearing impaired and deaf people.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published