C++ Inference using TFlite

TensorFlow Lite is an open source deep learning framework for on-device inference. On Android and Linux (including Raspberry Pi) platforms, we can run inferences using TensorFlow Lite APIs available in C++. The repository TensorFlowTTS and TensorFlow Lite help developers run popular text-to-speech (TTS) models on mobile, embedded, and IoT devices.

A demo of English or Mandarin TTS is available for linux platform (C++ Inference Demo). The pretrained models to be converted are download from the colab notebook (English or Mandarin). Mel-generator and Vocoder select FastSpeech2 and Multiband-MelGAN, respectively.

TFlite model convert method see colab notebook.

Results

• Comparison before and after conversion (English TTS)

  "Bill got in the habit of asking himself “Is that thought true?” \ 
  And if he wasn’t absolutely certain it was, he just let it go."
  

• Before conversion (Python)

  

• After conversion (C++)

  

• Adding #3 in chinese text will create pause prosody in audio

这是一个开源的端到端中文语音合成系统"

"这是一个开源的#3端到端#3中文语音合成系统"