jiangzhuo9357/qwen3-tts-12hz-0-6b-customvoice-gguf

Correct the CUDA placement claim: only the codec moves; the talker and code predictor never leave the CPU

Regenerate the card so it matches the generator exactly

Re-cut every package, and add Q8_MIXED

Qwen3-TTS 12Hz 0.6B CustomVoice, converted for synthesize.cpp

initial commit