Join the conversation

Join the community of Machine Learners and AI enthusiasts.

Sign Up
OzTianluΒ 
posted an update 3 days ago
Post
83
zixi-51M: Not a stock voice. A voice. 🐾

NoesisLab/zixi-51m
Everyone is shipping 500B-parameter models. I shipped 51 million. With one voice. On purpose. 🎀

What: open-weights Chinese TTS. VITS2 lineage. 51.43M params. Runs on your laptop CPU β€” no cloud, no API key, no GPU bill, no meter running.πŸ’Έ

The voice: not a stock voice, not a 5,000-speaker average. One real human's voice β€” mine. Timbre, rhythm, breath included. The sample on themodel card is one take, default parameters, recorded by absolutely no one. 🎧

What's NOT inside:

β€’ ❌ BERT (zero. the text encoder vibes on phones + tones alone)
β€’ ❌ language routing (Chinese-exclusive: 77 symbols, 6 tones, one job)
β€’ ❌ my training data (the corpus is my own writing and stays mine β€” open weights, closed corpus, that's the deal 🀝)

The trick: embedding transplant. Phone, tone and speaker rows inherited from the MeloTTS-Chinese base by symbol name β€” no random init. Themodel woke up already knowing Chinese. 🧠

Built solo, corpus to packaging, by a writer who is also a violinist. Prosody is my day job. 🎻

Use it, fork it, fine-tune your own voice. Just don't put words in a real person's mouth. πŸ™

πŸ”— Weights: NoesisLab/zixi-51m
πŸ’» Code: https://github.com/lizixi-0x2F/zixi-51M
In this post