ZeroTTS

Vietnamese Zero-Shot Text-to-Speech (TTS) with real-time streaming and voice cloning from seconds of audio. Fast, natural, and optimised for CPU inference.

The most accurate open Vietnamese TTS we know of — 4× fewer word errors than the next best model, and it runs faster than real time on a laptop CPU.

• 🎯 Ultra-natural — 2.91 UTMOS above every other open Vietnamese system, with near-zero dead air (0.029 s).

• 🗣️ Zero-shot voice cloning — a voice is a small latent array; drop it in and the model speaks in it, cloned from as little as 3 seconds of reference audio (up to 30 seconds). No fine-tuning, no per-speaker training.

• ⚡ Real-time on CPU, streaming — ~2× faster than real time (RTF 0.5×), first audio chunk in ~70 ms. No GPU required.

• 🇻🇳 Built for Vietnamese — tones, code-switched English, and reads 31/12/2025 and ZeroTTS without text normalizer.

• Code, examples, browser demo: https://github.com/zeroweight-ai/ZeroTTS

• Benchmark dataset: https://huggingface.co/datasets/zeroweight-ai/ZeroBench-TTS

• Blogpost: https://zeroweight.ai/blog/zero-tts

Samples

Two-speaker conversation

Long-form narration

News read, code-switched English

Cross-lingual

Reference audio (Vietnamese)

Output (English)

Usage

pip install zerotts

from zerotts import ZeroTTS

tts = ZeroTTS.from_pretrained("zeroweight-ai/ZeroTTS")
audio = tts.synthesize("Xin chào các bạn, mình là ZeroTTS.", voice="maichi")
tts.save_audio(audio, "out.wav")

Streaming, with first audio in roughly 70 ms:

import queue

import numpy as np
import sounddevice as sd   # pip install sounddevice

TEXT = ("Đây là chế độ phát trực tuyến. Âm thanh được tạo ra và phát ngay lập tức, "
        "không cần chờ toàn bộ đoạn văn hoàn thành. Nhờ vậy, người nghe chỉ mất "
        "khoảng 70 mili giây là đã nghe thấy câu đầu tiên, ngay cả khi mô "
        "hình đang chạy trên CPU của một chiếc laptop bình thường.")

pending, tail = queue.Queue(), np.zeros(0, dtype="float32")

def feed(outdata, frames, _time, _status):
    global tail
    while len(tail) < frames and not pending.empty():
        tail = np.concatenate([tail, pending.get_nowait()])
    n = min(frames, len(tail))
    outdata[:n, 0] = tail[:n]
    outdata[n:] = 0
    tail = tail[n:]

with sd.OutputStream(samplerate=tts.sample_rate, channels=1,
                     dtype="float32", callback=feed):
    for chunk in tts.synthesize_stream(TEXT, voice="maichi"):
        pending.put(chunk.reshape(-1))   # chunk is (1, n) float32 at 48 kHz
    while not pending.empty() or len(tail):
        sd.sleep(50)                     # let the buffer drain before closing

Benchmarks

Measured on ZeroBench-TTS

Every system reads raw text — dates, numbers and acronyms verbatim, exactly as they appear in the wild, with no text frontend in front of the model.

4× fewer word errors than the next-best system, and the fastest of the four on CPU. The gap is much wider in latency than in throughput: the two XTTS fine-tunes also beat real time (0.71×) but need seconds to emit their first sample, while OmniVoice is 6× slower than real time. All three are sized and tuned for a GPU, and it shows.

Full comparison tables, per-subset breakdowns, and CPU speed methodology: docs/BENCHMARKS.md

Speed — CPU

RTF (realtime factor, wall-clock synthesis time ÷ output audio duration — lower is faster; below 1× is faster than real time) and time-to-first-audio, all measured on CPU, single request, 8 inference threads pinned to a dedicated core pool (no other synthesis running concurrently). Three Vietnamese samples — short (26 chars), medium (77 chars), long (227 chars) — each run 6 times with the first 2 (cold-cache) discarded; figures below are the mean of the remaining 4.

ZeroTTS's time-to-first-audio comes from its real streaming path — first audio frame, not first full utterance. The three baselines have no working CPU streaming path, so their TTFA is the time to the complete utterance.

Voices, and voice cloning

A voice is a small array of speaker latents, (1, n_voice_queries, d_model), shipped as a .npz under voices/. That array is the entire speaker conditioning — no reference transcript, no audio prompt.

Voice cloning is not available in this release. Those latents come from a voice encoder that reads a reference clip, and that encoder is not published. This repository ships ready-to-use voices; it cannot create new ones from audio.

To get latents for your own speaker, see zeroweight.ai or get in touch.

Because a voice is just an array, latents obtained that way drop into voices//voice.npz and work with no code change.

Intended use and limitations

Built for Vietnamese. It handles English words embedded in Vietnamese text (code_switch), but it