NatiQ: An End-to-end Text-to-Speech System for Arabic

TL;DR

NatiQ is an end-to-end Arabic text-to-speech system utilizing encoder-decoder architectures and vocoders, achieving high naturalness scores and demonstrating the effectiveness of transformer-based models for speech synthesis.

Contribution

The paper introduces NatiQ, a novel end-to-end Arabic TTS system combining multiple models and vocoders, with new results on speech quality and efficiency.

Findings

Achieved MOS scores of 4.21 and 4.40 for two voices.

End-to-end ESPnet model outperformed other architectures in objective metrics.

System demonstrates real-time speech synthesis capability.

Abstract

NatiQ is end-to-end text-to-speech system for Arabic. Our speech synthesizer uses an encoder-decoder architecture with attention. We used both tacotron-based models (tacotron-1 and tacotron-2) and the faster transformer model for generating mel-spectrograms from characters. We concatenated Tacotron1 with the WaveRNN vocoder, Tacotron2 with the WaveGlow vocoder and ESPnet transformer with the parallel wavegan vocoder to synthesize waveforms from the spectrograms. We used in-house speech data for two voices: 1) neutral male "Hamza"- narrating general content and news, and 2) expressive female "Amina"- narrating children story books to train our models. Our best systems achieve an average Mean Opinion Score (MOS) of 4.21 and 4.40 for Amina and Hamza respectively. The objective evaluation of the systems using word and character error rate (WER and CER) as well as the response time measured by real-time factor favored the end-to-end architecture ESPnet. NatiQ demo is available on-line at https://tts.qcri.org