🗣 Easy text-to-speech for your Gatsby site, powered by Amazon Polly.


⚠️ This project is under heavy development, so breaking changes may occur on our road to a stable v1.0.0. Any bug reports and contributions will be highly appreciated.

Pull requests are welcome! npm GitHub license


Check out the ▶️ LIVE DEMO!

Demo Screencast

Also check out the example project repository!


npm install --save gatsby-mdx-tts

How to use


  1. In order to use this plugin you need an AWS account. You can use the text-to-speech service (AWS Polly) for free for the first 12 months (up to a couple million words to be precise).

    Attention: If you exceed the limits or use it after your initial free tier, using this plugin will generate costs in your AWS account!

  2. As this is a plugin for gatsby-plugin-mdx it will only work if you have that plugin installed and configured properly as well.

Mandatory configurations


To include the plugin just add it to your gatsby-plugin-mdx configuration in the gatsbyRemarkPlugin section. In case you have multiple gatsbyRemarkPlugins configured is very important that you put the gatsby-mdx-tts plugin to first position!

Also, you need to include a couple of mandatory configurations:

// In your gatsby-config.js
plugins: [
    resolve: `gatsby-plugin-mdx`,
    options: {
      gatsbyRemarkPlugins: [
          resolve: "gatsby-mdx-tts",
          options: {
            awsRegion: "us-east-1",
            defaultVoiceId: "Justin",

AWS credentials

The plugin requires your AWS credentials in order to generate the text-to-speech files.

There are two ways to configure your AWS credentials:

  1. (recommended) The recommended way is to create a shared credentials file. You probably already have one if you used the AWS CLI before.
  2. To override the credentials defined in a shared credentials file or to easily build on a CI environment you can optionally pass in the AWS credentials using plugin configuration options:
// In your gatsby-config.js
  resolve: "gatsby-mdx-tts",
  options: {
    awsCredentials: {
      accessKeyId: process.env.GATSBY_AWS_ACCESS_KEY_ID,
      secretAccessKey: process.env.GATSBY_AWS_SECRET_ACCESS_KEY,

Attention: If you choose to go with option 2 it is highly recommended to work with environment variables (as seen in the example above)! Do not directly paste your AWS credentials into your gatsby-config.js file and commit it to git as this would be a security issue!

All configurations

Option Required Example
awsRegion Yes "us-east-1"
defaultVoiceId Yes "Justin"
awsCredentials No { "accessKeyId": process.env.GATSBY_AWS_ACCESS_KEY_ID, "secretAccessKey": process.env.GATSBY_AWS_SECRET_ACCESS_KEY }
defaultSsmlTags No "<prosody rate='70%'>$SPEECH_OUTPUT_TEXT</prosody>"
defaultLexiconNames No ["LexA", "LexB"]
ignoredCharactersRegex No /·/
speechOutputComponentNames No ["CustomComponent"]
About defaultSsmlTags:
  • For an overview of all supported SSML tags check out the supported SSML tags list in the AWS docs.
  • The surrounding <speak> tag is added automatically.
  • The variable $SPEECH_OUTPUT_TEXT will be replaced with the speech output text.
About ignoredCharactersRegex:

If your text contains special characters that should not be vocalized (e.g. fear·ful should be just read as fearful) you can use the ignoredCharactersRegex to define the characters to be ignored.

You might also want those words not to be split up during word marking. Therefore also check out Ignore word splitting characters.

About speechOutputComponentNames:

If you want to use your own component to handle the generated speech output you can specify its name using the speechOutputComponentNames option. The plugin will then use this instead of SpeechOutput to extract the text to be used for TTS generation. It is also possible to define multiple component names. Like that you can customize the way speech output is handled. Find more information about this in the customization chapter.

Embed speech output in your MDX

After configuring the plugin you can now add the <SpeechOutput></SpeechOutput> component to your MDX files. The surrounded content will then be playable. You can add multiple speech output blocks to your content, but make sure the id is always set to an unique value over all occurrences. Also, it is important that there is an empty line between the SpeechOutput tags and the content to get it working.

import SpeechOutput from "gatsby-mdx-tts/SpeechOutput"

This text will be outside the speech output.

<SpeechOutput id="inside">

But this text will be playable. Please consider that:

- The play button is added automatically.
- The words in this text are marked one by one during text output.



Define individual speech output parameters

To define speech output parameters for individual <SpeechOutput> components you can pass them as props. This will override the eventually configured default parameters.

Prop Required Example
id Yes "my-individual-speech-output"
lexiconNames No ['LexA', 'LexB']
ssmlTags No "<prosody rate='70%'>$SPEECH_OUTPUT_TEXT</prosody>"
voiceId No "Hans"


<SpeechOutput voiceId="Hans" ssmlTags="<prosody rate='70%'>\$SPEECH_OUTPUT_TEXT</prosody>" id="my-individual-speech-output" lexiconNames={['LexA', 'LexB']}>

As you can see the order of the props does not matter. However, it is important to pass the props in the correct types (e.g. do not pass the lexiconNames as a string but as an array).

Play button

To customize the play button you can use the optional SpeechOutput component prop customPlayButton. Just pass in your custom play button component.

If you choose to use a custom play button component, make sure it uses the PlayButtonProps exported from this plugin.

Speech output handling

You can replace the whole speech output handling by using your own React component instead of the default SpeechOutput component. Like that, the TTS files are still generated during build phase but you can then do whatever you want with those files inside your component at runtime. To do so, use the speechOutputComponentNames configuration option (see About speechOutputComponentNames).

If you choose to use your own component, make sure it uses the SpeechOutputProps exported from this plugin.

Custom useSound hook

In case you would like to manage playing sounds by yourself you can pass an optional hook to the useCustomSoundHook prop of the SpeechOutput component. It has to follow the UseSoundHookSignature type as exported from UseSound.ts (which is the default sound hook).

Ignore word splitting characters

You might use characters that split a word into two, e.g. fear·ful. Those word parts are now marked individually by default. To avoid this, you can define the characters to ignore with the ignoredWordSplittingCharactersRegex prop.

Probably you also don’t want this character to be vocalized during speech output. Therefore make sure you also configure the ignoredCharactersRegex in the plugin options.

Event listeners

To be able to react to certain events you can register the following event listeners:


When a speech output is played the spoken words are highlighted in the text simultaneously. The onWordMarked listener is called as soon as a new word is highlighted and delivers the currently highlighted word as a string. When no word is highlighted (anymore) the string is empty.

Contribute 🦸

Contributions are more than welcome! I would love to see text-to-speech becoming a thing in the already very accessible Gatsby ecosystem. If you agree with this and would like to join me on this mission it would be awesome to get in touch! 😊

Please feel free to create, comment and of course solve some of the issues. To get started you can also go for the easier issues marked with the good first issue label if you like.


The MIT License


The gatsby-mdx-tts library is maintained and sponsored by the Swiss web and mobile app development company Florian Gyger Software.