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

Moss Tts

MOSS-TTS is an open-source Python model family for generating speech and sound, with models for long-form narration, multi-speaker dialogue, voice design, environmental effects and streaming text-to-speech.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Moss 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 the project when selecting or integrating an open model for a defined speech or sound-generation task; consult the model-specific resources because the family contains variants aimed at different workloads.

Who it is for

  • Developers evaluating open speech and audio generation models.
  • Teams creating narrated content, multi-speaker dialogue, dubbing or podcasts.
  • Builders exploring voice design, cloning or generated environmental audio.

Documented uses

  • Turn text into long-form spoken narration.
  • Generate speech for multiple dialogue speakers.
  • Create designed voice or character speech.
  • Produce environmental sound effects from text.
  • Add streaming speech generation to an application.
Before you choose

Requirements & limits

The repository is public and identifies the project as open source under the Apache-2.0 license. It links to model weights, an AI Studio and API documentation; these are distinct routes, and the repository information does not specify their account or pricing terms.

  • Model variants target different jobs: the README points to MOSS-TTSD for multi-speaker dialogue, MOSS-TTS-Realtime for streaming, and v1.5 for multilingual long-form narration and cloning.
  • The README lists MOSS-TTS-Nano for speech and voice cloning on CPU or in a browser; this is a model-specific positioning, not a general compatibility guarantee.
  • Python is the repository's primary language, and the project links to model weights, fine-tuning material and serving-backend resources.
  • The model family is licensed Apache-2.0.

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

Set up the MOSS-TTS Python environment with Conda

Prepare an isolated Python 3.12 environment and install the MOSS-TTS runtime dependencies.

Before you start

  • Conda
  • FFmpeg: install with apt on Debian/Ubuntu or Homebrew on macOS
  • A CUDA 12.8-compatible PyTorch installation path as specified by the dependency command

Steps

  1. Create and activate the environment: conda create -n moss-tts python=3.12 -y then conda activate moss-tts.

  2. Install FFmpeg with sudo apt-get install -y ffmpeg on Debian/Ubuntu, or brew install ffmpeg on macOS.

  3. Clone the repository and enter it: the command below then cd MOSS-TTS.

    Documented command
    git clone https://github.com/OpenMOSS/MOSS-TTS.git

  4. Install the runtime dependencies: the command below.

    Documented command
    pip install --extra-index-url https://download.pytorch.org/whl/cu128 -e ".[torch-runtime]"

Expected outcome

An isolated Python environment with the MOSS-TTS runtime dependencies installed.

Guide limitations

  • The project recommends Transformers 5.0.0 in a clean isolated environment.
  • The documented pip install command targets CUDA 12.8; the README notes that other backends may be used by changing the backend, but does not provide the equivalent pip command here.
  • FFmpeg is required by torchcodec for audio I/O.
  • FlashAttention 2 is optional and only available on supported GPUs.
Documentation 1 ↗
Source-based practical guide

Set up MOSS-TTS with uv

Create an isolated Python 3.12 environment and install MOSS-TTS runtime dependencies with uv.

Before you start

  • uv installed
  • A CUDA 12.8-compatible PyTorch installation path as specified by the dependency command

Steps

  1. Clone the repository and enter it: the command below then cd MOSS-TTS.

    Documented command
    git clone https://github.com/OpenMOSS/MOSS-TTS.git

  2. Create and activate the virtual environment: the command below then source .venv/bin/activate.

    Documented command
    uv venv --python 3.12 .venv

  3. Install runtime dependencies: the command below.

    Documented command
    uv pip install --torch-backend cu128 -e ".[torch-runtime]"

Expected outcome

A uv virtual environment with the MOSS-TTS runtime dependencies installed.

Guide limitations

  • The README recommends a clean isolated Python environment with Transformers 5.0.0.
  • The shown install command selects the CUDA 12.8 backend; the README gives cpu and cu126 as examples of alternative backend names.
  • FlashAttention 2 is optional, requires a supported GPU, and may be skipped if its build fails.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
7 October 2026
Original website title
OpenMOSS/MOSS-TTS · Official homepage title · retained in its source language
How we review sourcesSuggest a correction