Voice vocoder speed vs quality: HiFi-GAN vs BigVGAN, 48 kHz for +0.2 MOS
HiFi-GAN vs BigVGAN: isolating activation and anti-aliasing yields +0.2 MOS at 48 kHz, but only when mel inputs, hardware, and batch size match. Independent technical note mirroring the canonical version: https://clonemyvoice.io/blog/voice-vocoder-speed-vs-quality-hifi-gan-vs-bigvgan-48-khz-for-02-mos.php
Authors
- Dylan Cooper
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-10
- DOI
- https://doi.org/10.5281/zenodo.24101573
- Primary Topic
- Speech Recognition and Synthesis
- Type
- preprint