OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Audio & Voice AI

CONFUCIUS4 Tts

CONFUCIUS4 Tts is a multilingual, cross-lingual text-to-speech model that uses a speech encoder and a large language model to generate speech from text while preserving a reference speaker's identity across languages.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

CONFUCIUS4 Tts — OpenViST project profile in Audio & Voice AI.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when text needs to become speech in multiple languages and the generated voice should retain the identity of a supplied reference speaker.

Who it is for

  • Developers prototyping multilingual text-to-speech applications.
  • Users creating narration in multiple languages from a reference voice.
  • Researchers exploring zero-shot and cross-lingual voice transfer.

Documented uses

  • Generate speech from text in one of the 14 listed languages.
  • Transfer a reference speaker's voice identity across languages.
  • Try the system through its linked online demo.
  • Run inference from Python or the provided command-line example.
Before you choose

Requirements & limits

The repository is public and identifies Apache 2.0 as its code license. It links to an online demo; local setup instructions specify Python 3.10 and CUDA 12.6.

  • The repository lists 14 supported languages: Chinese, English, Japanese, Korean, German, French, Spanish, Indonesian, Italian, Thai, Portuguese, Russian, Malay and Vietnamese.
  • The basic inference path uses Hugging Face Transformers; an alternative vLLM path specifies vLLM 0.16.0 with the V1 engine.
  • The vLLM instructions warn that GPU architecture, driver and CUDA requirements can affect whether it runs on a given machine.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Run zero-shot speech synthesis locally

Generate a WAV speech file from text using a reference WAV and the repository's example script.

Before you start

  • Python 3.10 and CUDA 12.6, as listed by the repository.
  • A cloned Confucius4-TTS repository with dependencies installed from requirements.txt.
  • A reference WAV file and text in a supported language.

Steps

  1. Clone the repository and change into its directory.

  2. Create and activate a conda environment with Python 3.10.

  3. Install dependencies with pip install -r requirements.txt.

  4. Run example.py with --prompt_wav, --text, --lang, --out and --config config/inference_config.yaml.

Expected outcome

The example command is configured to write generated speech to the specified WAV output path.

Guide limitations

  • The listed setup requirements include CUDA 12.6.
  • The guide documents the repository's example command; it does not establish compatibility with other environments or hardware.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
6 November 2026
Original website title
netease-youdao/Confucius4-TTS · Official homepage title · retained in its source language
How we review sourcesSuggest a correction