Blog Post

Azure AI Foundry Blog
8 MIN READ

Azure Neural TTS upgraded with HiFiNet, achieving higher audio fidelity and faster synthesis speed

QinyingLiao's avatar
QinyingLiao
Icon for Microsoft rankMicrosoft
Nov 03, 2020
This post was co-authored with Jinzhu Li and Sheng Zhao   Neural Text to Speech (Neural TTS), a powerful speech synthesis capability of Cognitive Services on Azure, enables you to convert text to...
Vocoder-Training.png
Updated Nov 27, 2024
Version 6.0
azure ai services
Neural TTS
speech
Jinzhu_Li's avatar
Jinzhu_Li
Icon for Microsoft rankMicrosoft
Feb 15, 2022

Overall, they are both GAN based vocoder, but their generator and discriminator are different. HiFiNet's inference speed is also much faster, about 3x of hifigan smallest model(v3).