techygrrrl
techygrrrl's blog

techygrrrl's blog

Multilingual Text-to-Speech (TTS) app for Twitch on macOS

Multilingual Text-to-Speech (TTS) app for Twitch on macOS

techygrrrl's photo
techygrrrl
ยทJun 7, 2022ยท

4 min read

Table of contents

  • Features
  • Download and Install (soon ๐Ÿคž)
  • Watch me build it
  • Get some code

Previously, I had built a text-to-speech app on Windows to help me listen to my Twitch chat while streaming. You can read about the project, why I built it, and download it here. This post is about the macOS version which I built on stream. If you're curious about my streams, follow me on Twitch @techygrrrl.

Screen Shot 2022-06-06 at 8.41.02 PM.png

Features

Multilingual

This text-to-speech app supports all languages that Apple supports by default.

Here's a snippet of code that shows which languages are supported:

.init(name: "Arabic", languageKey: "ar"),
.init(name: "Chinese", languageKey: "zh"),
.init(name: "Czech", languageKey: "cs"),
.init(name: "Danish", languageKey: "da"),
.init(name: "Dutch", languageKey: "nl"),
.init(name: "English", languageKey: "en"),
.init(name: "Finnish", languageKey: "fi"),
.init(name: "French", languageKey: "fr"),
.init(name: "German", languageKey: "de"),
.init(name: "Greek", languageKey: "el"),
.init(name: "Hebrew", languageKey: "he"),
.init(name: "Hindi", languageKey: "hi"),
.init(name: "Hungarian", languageKey: "hu"),
.init(name: "Indonesian", languageKey: "id"),
.init(name: "Italian", languageKey: "it"),
.init(name: "Japanese", languageKey: "ja"),
.init(name: "Korean", languageKey: "ko"),
.init(name: "Norwegian", languageKey: "nb"),
.init(name: "Polish", languageKey: "pl"),
.init(name: "Portuguese", languageKey: "pt"),
.init(name: "Romanian", languageKey: "ro"),
.init(name: "Russian", languageKey: "ru"),
.init(name: "Slovak", languageKey: "sk"),
.init(name: "Spanish", languageKey: "es"),
.init(name: "Swedish", languageKey: "sv"),
.init(name: "Thai", languageKey: "th"),
.init(name: "Turkish", languageKey: "tr")

Support depends on the macOS version and installed voices. At time of development, on macOS 11, the above languages are supported. Please note that not all of them have been tested.

Screen Shot 2022-06-06 at 8.39.52 PM.png

Speed, Volume and Stop Controls

You are able to configure the speed and volume controls on the Listen screen. The recommended speed is 0.5.

You can also stop the app from speaking if it's in the middle of speaking. Note that any already-queued up messages will not be spoken. Once another person sends a message after it's been stopped, they will continue to be read aloud. To stop it entirely, quit the app or lower the volume to 0%.

Screen Shot 2022-06-06 at 8.38.58 PM.png

Noise reduction features

The app includes features to reduce noise from official Twitch emotes. BTTV emotes and unicode emojis not currently supported so they will be read aloud.

Word Replacements

This app has the ability to replace words in text, just like the Windows version I built. It runs on the message text and the username.

This can be useful for the following reasons:

  • expand internet slang / short forms
  • override username pronunciation

Screen Shot 2022-06-06 at 8.39.07 PM.png

Ignore List

This app adds the ability to ignore usernames. This is useful if you do not want to hear your bots' messages being read aloud.

Screen Shot 2022-06-06 at 8.39.31 PM.png

Download and Install (soon ๐Ÿคž)

Once I pay for an Apple developer account ($100/yr) to sign and distribute the app, it will become available. Unfortunately, Apple does not allow developers to distribute an app without paying the annual fee. If you need access to a text-to-speech app and have a Windows computer, you can download and install the Windows version. It's available now and is free.

image.png

Watch me build it

I built this in 7 streams. Streams ranged from 5 to 9 hours long. It was an epic journey learning SwiftUI and the text-to-speech API's on macOS.

You can watch my journey learning SwiftUI and macOS development on Twitch in the following video collection: twitch.tv/collections/-9v-a8uO8RbfCA

You can also see individual videos:

This is what it looked like on day 1 before the stream started:

image.png

Get some code

Code up to day 4 is available to patrons on Patreon (all membership tiers). It's not the final product but contains the core functionality:

  • detects language and reads out the message in that detected language
  • change volume
  • change speed
  • word replacement
  • ignore usernames

What's missing is the ability to configure which languages you want to listen to (all languages on your computer will be used), as well as the pretty UI.

Become a patron here: patreon.com/techygrrrl

Here's a screenshot of the app on day 4.

image.png

I plan on publishing via the mac App Store once I get an Apple developer account so people can have access. It will either be free or offered at a small fee in order to recover costs paid to Apple.

If you want a free text-to-speech app now and you have Windows, you can get my Windows one here. There's also a very quick and dirty one in the browser that is English only and randomizes the voices: lrvbfs.csb.app

ย 
Share this