README.md

Mozilla TTS-TFLite

Docker image for Mozilla TTS with Tensorflow Lite support based on @erogol's TFLite example and edited from synesthesiam/docker-mozillatts.

Includes TFLite-optimized versions of @erogol's pre-built LJSpeech Tacotron2 English model and Multiband MelGAN vocoder. See below for links to specific checkpoints.

Using

$ docker build -t tts .
$ docker run -it -p 5002:5002 tts

Visit http://localhost:5002 for web interface.

Do HTTP GET at http://localhost:5002/api/tts?text=your%20sentence to get WAV audio back:

$ curl -G --output - \
    --data-urlencode 'text=Welcome to the world of speech synthesis!' \
    'http://localhost:5002/api/tts' | \
    aplay

Building Yourself

The Docker image is built using the TFLite Example. You'll need to manually download the model and vocoder checkpoints/configs:

• model/tts_model.tflite
• model/config.json
• vocoder/vocoder_model.tflite
• vocoder/config_vocoder.json
• scale_stats.npy