Mimic 3 logo

Mimic 3

Mimic 3 - Privacy-focused neural TTS engine | Product Hunt

Mimic 3 is a privacy-focused open-source neural Text to Speech (TTS) engine.

In human terms that means it sounds great and can run completely offline on hardware you control. A cloud service is in the works for people who want an “easy button” or for hardware that can’t math fast enough.

Hear My Voices

Key Features

Run completely offline, even on embedded systems

Speak over 25 languages

Over 100 pre-trained voices available

Run multi-speaker models

Supports Speech Synthesis Markup Language

Custom word pronunciations

Mimic 3 can run faster than real-time, on resource constrained hardware like the Raspberry Pi 4, and completely offline. This is an absolute game-changing feature for a privacy-focused voice assistant like the Mycroft Mark II.

It can even be used as a drop-in replacement for MaryTTS, an engine used ubiquitously in self-hosted IoT systems like Home Assistant. There is also GPU acceleration support using CUDA.

You no longer have to choose between a natural sounding voice, or privacy. Mimic 3 sounds beautiful and won’t send your data anywhere.

Language Support

Mimic 3 already provides voices in 25 languages:

  • Afrikaans
  • Bengali
  • Dutch
  • English (UK)
  • English (US)
  • Farsi
  • Finish
  • French
  • German
  • Greek
  • Gujarati
  • Hausa
  • Hungarian
  • Italian
  • Javanese
  • Kiswahili
  • Korean
  • Nepali
  • Polish
  • Russian
  • Setswana
  • Spanish
  • Telugu
  • Ukrainian
  • Vietnamese
  • Yoruba

New languages are being added all the time. If you have a data set and are looking for help in training, please reach out to us.

Multi-Speaker Models

There are over 100 different voices already publicly available for Mimic 3. These are distributed as either single or multi-speaker models.

A multi-speaker model provides the ability to generate speech using any of the voices in that model without any overhead of loading a new model into memory. This enables rich multi-speaker interactions in your application by allowing you to switch voices on the fly.

Want more of the nerdy details? Check out our documentation.

Get Mimic 3

The easiest way to make use of Mimic 3 is to grab yourself a Mycroft Mark II which uses Mimic 3 as the standard TTS engine. Pre-orders for the first two production runs of the Mark II have sold out.

PRE-ORDER YOUR MARK II TODAY

Mimic 3 can also be used on any existing Mycroft installation with a Raspberry Pi 4 or better using our TTS plugin. Beyond Mycroft you can run Mimic 3 on Linux, or as a Docker container. It even includes its own web server.

See our documentation for all the options.

Coming Soon...

Mimic 3 Cloud

Our next Mimic 3 release will be a hosted cloud-service. Ideal for those who need speech synthesized for ultra-low powered devices, or that need even more customization and premium voices.

This will include a cloud-based service for commercial partners. Register your interest below.

What about Windows, Mac, Flatpak, Snap, AppImage, ...?

Over time we will be releasing Mimic 3 on more platforms, and in different formats. This might include Windows, Mac OS, Flatpak, Snap, AppImage, iOS or Android.

If you want to be the first to hear about new Mimic 3 releases, subscribe for product updates below.

Stay Up To Date

Register for updates to be the first to find out about future releases of Mimic 3.