Qwen3-TTS is a family of advanced multilingual, controllable, robust, and streaming text-to-speech models. Trained on over 5 million hours of speech data spanning 10 languages, Qwen3-TTS supports state-of-the-art 3-second voice cloning and description-based control.
ZONOS3-Mini-hf
Zyphra·released 11 Nov 2026
hf download Zyphra/ZONOS3-Mini-hfZONOS3-Mini is Zyphra's compact text-to-speech model supporting 43 languages, with about 100 million active parameters per token (800 million total in the backbone).
How much VRAM does ZONOS3-Mini-hf need?
At FP16 / BF16 ZONOS3-Mini-hf needs about 2.8 GB of GPU memory.
| Quantisation | Total |
|---|---|
| 4-bitRecommended default for local use | ≈1.2 GB |
| 8-bitLossless in practice, twice the size of Q4 | ≈1.7 GB |
| FP16 / BF16recommendedFull precision, as released | ≈2.8 GB |
Estimates for the weights plus runtime overhead; activations vary with resolution and batch size. Pick a GPU above and the fit column checks every row against it.
Run ZONOS3-Mini-hf locally
These commands load the original weights from Hugging Face.
from transformers import pipeline
pipe = pipeline("text-to-speech", model="Zyphra/ZONOS3-Mini-hf")
result = pipe("Self-hosting puts you in control of your data.")
print(result)About ZONOS3-Mini-hf
ZONOS3-Mini is Zyphra's compact text-to-speech model supporting 43 languages, with about 100 million active parameters per token (800 million total in the backbone). This repo packages it for plain transformers (remote code, trust_remote_code=True): the language model, the causal multi-resolution iSTFT codec decoder (keep2_plain) and the MECT-B2 speaker encoder in one checkpoint, with the SuperBPE tokenizer and the 27 release voices. It synthesizes 44.1 kHz mono speech on CPU or GPU with nothing but torch and transformers. Model details, samples and evaluations: the blog post and technical report. The native runtimes (CUDA, MLX, ggml, WebGPU) and the full HTTP API live in the ZONOS3-Mini repository. Try the browser studio, its Lite voice-cloning interface, or Zyphra Cloud. The same 43 languages as the main model, with the native runtimes' request codes (English has two locales): Text normalization uses NeMo grammars via zyphra_textnorm for English, Mandarin, Japanese, Korean, German, Spanish, French, Italian and Portuguese. Other languages pass through as written, with line breaks collapsed; they do not require the optional normalizer. language selects normalization, not a language token in the prompt.
Full model card on Hugging Face →Alternatives to ZONOS3-Mini-hf
All alternatives →Qwen3-TTS is a series of advanced multilingual, controllable, robust, and streaming text-to-speech models developed by the Qwen team.
VibeVoice-Realtime is a lightweight real‑time text-to-speech model supporting streaming text input and robust long-form speech generation.
OmniVoice is a massively multilingual zero-shot text-to-speech (TTS) model supporting over 600 languages. Built on a novel diffusion language model-style architecture, it delivers high-quality speech with superior inference speed, supporting voice cloning and voice design.
This repository contains Ming-3.0-tts-0.5B, the dense Ming-3.0-tts checkpoint built on Qwen2.5-0.5B. The checkpoint includes the backbone, DiT head, audio tokenizer, speaker encoder and text tokenizer.
Qwen3-TTS covers 10 major languages (Chinese, English, Japanese, Korean, German, French, Russian, Portuguese, Spanish, and Italian) as well as multiple dialectal voice profiles to meet global application needs.
ZONOS3-Mini-hf FAQ
- How much VRAM does ZONOS3-Mini-hf need?
- About 2.8 GB at FP16 / BF16.
- Can ZONOS3-Mini-hf run on the RTX 4060?
- Yes. It fits at FP16 / BF16 (about 2.8 GB). It fits comfortably on the Apple M4 (16 GB), Radeon RX 6700 XT, RTX 3060 12GB.
- Can I use ZONOS3-Mini-hf commercially?
- Yes. It is released under Apache 2.0. Permissive: use, modify and sell freely; keep the notice and licence text.
- How do I run ZONOS3-Mini-hf locally?
- It runs with Transformers. Each has copy-paste examples in its docs.
Rate this model
Data from Hugging Face, the model card and public leaderboards. Memory and speed figures are estimates; real numbers depend on the runtime, context length and drivers. Always check the licence yourself before commercial use.
0 comments