Narrator is a mod designed primarily for people with visual impairments. It reads aloud the text of various game elements, as many ordinary screen readers do. However, Narrator does this directly in Minecraft and when you hover the cursor (in 95% of cases), which is much more convenient than constantly pressing Ctrl+C/Ctrl+V. **How does it work?** When you hover over a supported game element (such as a button), Narrator retrieves its text and passes it to the PiperTTS* subprocess. Piper converts the text into an audio file, which is immediately played in the game. *A third-party [text-to-speech engine](https://github.com/rhasspy/piper/tree/windows) that works with various voice models. **What steps need to be taken for Narrator to work?** 1. All the files required for PiperTTS are already included in the mod. You just need to launch Minecraft, and Narrator will automatically extract them to the _config/Narrator/PiperTTS_ folder. 2. Next, you need to download the voice model* for the current Minecraft client language and place it in the language folder generated at startup. For example, for English, the path will be _config/Narrator/PiperTTS/voices/en_US_ 3. That's it, you don't need to do anything else. You don't even need to restart the game; you can change the voice model at any time and it will work immediately. *Since there are quite a few voice models for each individual language, and they are quite large, Narrator does not arbitrarily attempt to impose a specific model on you. You can choose the model you like and use it for voiceovers. Here, you can check how each voice model will sound in practice: https://rhasspy.github.io/piper-samples/ And here you can download the selected model for the current client language: https://huggingface.co/rhasspy/piper-voices/tree/main **Important!** You only need two model files — one with the **.onnx** extension and the other with the **.onnx.json** extension. For each model, these files must always have the same name but different extensions. **Warning!** Do not use multi-speaker models, as they are not intended for use in Minecraft. Even if everything works fine with them, the quality of an individual speaker may be less than a single speaker model.

Arabic language support

Due to the peculiarities of working with the Arabic language, additional actions are required from the player to support it: 1. You need to download the [piper_windows_amd64.zip](https://github.com/rhasspy/piper/releases/download/2023.11.14-2/piper_windows_amd64.zip) archive (OR any [other](https://github.com/rhasspy/piper/releases/tag/2023.11.14-2)) and copy the **libtashkeel_model.ort** file from it to the _config/Narrator/PiperTTS_ folder (**not in voices**). 2. Also, the [voice model for Arabic](https://huggingface.co/rhasspy/piper-voices/tree/main/ar/ar_JO) has the language code ar_JO, unlike the Minecraft client language and the folder where you need to place the model: _config/Narrator/PiperTTS/voices/ar_SA_ 3. Once you have completed the previous two steps, everything should work as usual.

What can Narrator already narrate?

When clicked: – Chat messages (only the line you clicked, because right now I can't distinguish the lines of one message from another when they appear quickly in the chat, causing the voiceover to be played for multiple messages at once. In the future, I'll try to add voiceover for multi-line messages, but for now, I'll have to click each line manually) When hovered over: – Chat message tooltips (achievements, give, plain text) – Items and their tooltips – Inventories (blocks, mobs, and player) – Creative tabs – Achievements – Statistics – Book and Quill – Boss bars – Experience level – Potion effects – Debug mode (F3) – Splashes – Buttons and their tooltips – Mod List – Mod configurations – Most game screens – Input fields (text, numbers) – Description of single worlds – Description of multiplayer/LAN servers (including motd, online, ping, platform) When scrolling: – Selected item on the hotbar Automatically: – Hint on how to get off transport – Name of the playing record – End Poem

What is not currently supported?

– Profiler in debug mode (Shift+F3). I am not yet able to teach Narrator to correctly identify all profiler layers and change their status – Screens when creating/loading a world, teleporting between worlds, and other screens that appear temporarily – Tab list – Screens that appear in certain situations that I simply forgot about or didn't know about – Absolutely no mods. If something works, it's a coincidence, not support. But it's all in the plans…

FAQ

**Q:** How can I check if the mod is working? **A:** If you haven't heard any voiceovers yet, there is a more reliable way. When you join the world, a message about the mod's status will appear. You can also use the **/narrator status** command. If you followed the instructions correctly, the message will be green. If you forgot some part of the voice model, it will be yellow. And if you didn't pay attention to the instructions at all, the message will be red. In any case, you won't get confused, and you'll know what needs to be fixed. **Q:** What operating systems are supported? **A:** Windows, Linux, and macOS. The necessary files will automatically be extracted to the _config/Narrator/PiperTTS_ folder when the client is launched. **Q:** Does Narrator require an Internet connection to work? **A:** No. Narrator is completely autonomous and works without an Internet connection. **Q:** Can Narrator break anything in my game world, computer, or simply be considered cheating? **A:** No. Narrator is completely safe for your world, as it is entirely a client-side mod. It also does not do anything to your computer that Minecraft itself, your browser, or any other software that writes and reads data does not do. And it cannot be classified as a cheat, as it does not give the user any advantage whatsoever. It cannot amplify footstep sounds, show the location of mobs using echolocation, or do anything else that falls outside the definition of “text-to-speech”. **Q:** How can I test the selected voice model directly in the game? **A:** You can use the **/narrator test** command and write any text after it that will be voiced. **Q:** How do I work with the cache if I want to find out its total size or clear it? **A:** You can use the **/narrator cache info** command to check the size and number of files in the cache, and the **/narrator cache clear** command to completely clear the folder. **Q:** How can I adjust the sound volume? **A:** Narrator volume is linked to the Master Volume in the Music & Sounds settings. **Q:** Does the mod have a config? **A:** Yes. Narrator has a fairly comprehensive configuration file with clear parameter descriptions. You can customize all the mod's features to suit your modpack. **Q:** If I don't want to listen to a long text, what should I do? **A:** You can hover your cursor over any short text, such as the Done button, and the long text will be interrupted. But there is a better option. You can bind the Stop Narrator hotkey, which interrupts any narration when pressed. **Q:** What if I want to turn off Narrator, but only until I restart, or turn it off and turn it back on after a while? **A:** You can bind the Enable/Disable Narrator hotkey, which turns Narrator on and off for the current game session. **Q:** I found something that isn't narrated at the moment / I have other questions. What should I do? **A:** Let me know in the comments or, better yet, on our [Discord server](https://discord.gg/dqQXZHq).

**LEGAL DISCLAIMER: I have no affiliation with PiperTTS or Piper voice models. I do not own any rights to PiperTTS or Piper voice models. I only use them in accordance with their licenses and express my deepest gratitude to everyone involved in the development of PiperTTS and the Piper voice models.**

Narrator

Links

Featured versions

Project members

Details