AIHacker Noonabout 8 hours ago
Audio8 TTS 0.1B Clones Voices on Small GPUs
11 min read
Audio8's 0.1B text-to-speech model performs zero-shot voice cloning with 170M parameters, fitting consumer GPUs. It supports Chinese and English with 3.619 and 3.307 WER respectively. The model uses a dual-branch Falcon H1 architecture and operates at 44.1 kHz. Non-commercial licensing restricts deployment.
Level
Hype check
Tap to vote and see what everyone thinks.
Summary by ByteBrief
