Scaling Nvidia's Multi-Speaker Multi-Lingual TTS Systems With Zero-Shot TTS to Indic Languages
Lists
Only you see this
Abstract
In this paper, we describe the TTS models developed by NVIDIA for the MMITS-VC (Multi-speaker, Multi-lingual Indic TTS with Voice Cloning) 2024 Challenge. In Tracks 1 and 2, we utilize RAD-MMM [1] to perform few-shot TTS by training additionally on 5 minutes of target speaker data. In Track 3, we utilize P-Flow [2] to perform zero-shot TTS by training on the challenge dataset as well as external datasets. We use HiFi-GAN [3] vocoders for all submissions. RAD-MMM performs competitively on Tracks 1 and 2, while P-Flow ranks first on Track 3, with mean opinion score (MOS) 4.4 and speaker similarity score (SMOS) of 3.62.
BibTeX
@inproceedings{arora2024scaling,
author = {Akshit Arora and Rohan Badlani and Sungwon Kim and Rafael Valle and Bryan Catanzaro},
title = {{Scaling Nvidia's Multi-Speaker Multi-Lingual TTS Systems With Zero-Shot TTS to Indic Languages}},
booktitle = {Workshops @ ICASSP},
year = {2024},
url = {https://openreview.net/forum?id=gXoqMl0K4J},
}