audio8

Reproducible text-to-speech experiments in Castilian Spanish: voice design, phonetic accuracy, long-form stability and shrill-voice diagnosis for Qwen3-TTS, with Audio8 for comparison. Every result regenerates from one command.

Basque Text-to-Speech with Natural Prosody — fine-tuned from Audio8-TTS-0.6B

LoRA fine-tuning and adapter reload for Audio8 TTS on CUDA and Apple MPS

Wyoming Protocol TTS server for Audio8-TTS-Realistic-Air, ready for Home Assistant.

Local context-aware voice mimics for Lethal Company using Qwen3.5 and Audio8 TTS.

Generate multilingual speech and clone voices with this 0.6B-parameter zero-shot text-to-speech model.