Audio8: 170M-parameter open-source multilingual TTS model
CatAstro_Piyush · x · 2026-08-20
Audio8 TTS Preview 0.1B is announced as the world's smallest Transformer-based TTS model. Featuring approximately 170 million parameters, it supports multilingual speech synthesis and zero-shot voice cloning, delivering surprisingly strong, cloud-level quality in a dramatically smaller footprint.
More from Multimodal
- SDXL newb confused: Pony/Illustrious workflows make photorealism — which fine-tune to pick? — throwaway0204055 · 2026-08-20
- Photorealistic historical drama scene via Seedance 2.5 — SimplyAnnisa · 2026-08-20
- AI Narrative Music Video: My Cancer Story — VIV-AF-D · 2026-08-20
- AI Generated Video: No More Dust — marsplanet88 · 2026-08-20
- Krea 2 LoRA Training Guide: Character & Pose on 16GB VRAM — Opening-Knee-5913 · 2026-08-20
- Minimax H3 I2V Can't Keep White Background? User Seeks Solution — noaxxx2 · 2026-08-20