Speech Synthesis on LibriTTS
Metric: MCD (higher is better)
Metric: MCD (higher is better)
| # | Model↕ | MCD▼ | Extra Data | Paper | Date↕ | Code |
|---|---|---|---|---|---|---|
| 1 | WaveGlow | 2.3591 | No | WaveGlow: A Flow-based Generative Network for Sp... | 2018-10-31 | Code |
| 2 | SC-WaveRNN | 1.8854 | No | Speaker Conditional WaveRNN: Towards Universal N... | 2020-08-09 | Code |
| 3 | WaveFlow | 1.2455 | No | WaveFlow: A Compact Flow-based Model for Raw Audio | 2019-12-03 | Code |
| 4 | HiFi-GAN | 0.6603 | No | HiFi-GAN: Generative Adversarial Networks for Ef... | 2020-10-12 | Code |
| 5 | BigVGAN-base | 0.4564 | No | BigVGAN: A Universal Neural Vocoder with Large-S... | 2022-06-09 | Code |
| 6 | BigVSAN (w/ snakebeta) | 0.4129 | No | BigVSAN: Enhancing GAN-based Neural Vocoders wit... | 2023-09-06 | Code |
| 7 | BigVGAN | 0.3745 | No | BigVGAN: A Universal Neural Vocoder with Large-S... | 2022-06-09 | Code |
| 8 | BigVSAN | 0.3381 | No | BigVSAN: Enhancing GAN-based Neural Vocoders wit... | 2023-09-06 | Code |
| 9 | BigVGAN-v2 | 0.2903 | Yes | BigVGAN: A Universal Neural Vocoder with Large-S... | 2022-06-09 | Code |