Clone any voice from a 5-second sample. Speak 31 languages. Zero cloud. Zero subscription. Zero compromise.
# One command:
curl -fsSL https://xirtus.github.io/MOSSlanding/install-apt.sh | sudo bash
mosslandingOr manually:
wget https://github.com/xirtus/MOSSlanding/releases/latest/download/mosslanding_latest_amd64.deb
sudo dpkg -i mosslanding_latest_amd64.deb
mosslanding# Option 1: Download from GitHub Releases
wget https://github.com/xirtus/MOSSlanding/releases/latest/download/mosslanding-1.0.0-1-x86_64.pkg.tar.zst
sudo pacman -U mosslanding-1.0.0-1-x86_64.pkg.tar.zst
mosslanding
# Option 2: Build from AUR (coming soon — PKGBUILD in packaging/arch/)
# git clone https://aur.archlinux.org/mosslanding.git
# cd mosslanding && makepkg -si# Download DMG from Releases
open https://github.com/xirtus/MOSSlanding/releases/latest
# Drag Mosslanding.app to /Applications# Download and run installer from Releases
Invoke-WebRequest -Uri "https://github.com/xirtus/MOSSlanding/releases/latest/download/Mosslanding-Setup.exe" -OutFile "Mosslanding-Setup.exe"
.\Mosslanding-Setup.exe| Format | System | Install Command | Status |
|---|---|---|---|
.deb |
Debian / Ubuntu / Mint / Pop!_OS | sudo dpkg -i mosslanding_*.deb |
✅ Available |
.pkg.tar.zst |
Arch Linux / Manjaro / EndeavourOS | sudo pacman -U mosslanding-*.pkg.tar.zst |
✅ Available |
.dmg |
macOS (Apple Silicon) | Download & drag to Applications | ✅ Available |
.exe / .ps1 |
Windows 10/11 | Run installer | ✅ Available |
.AppImage |
Any Linux | Download & run | ✅ Available |
| APT Repo | Debian/Ubuntu | sudo apt install mosslanding |
🚧 Setup required |
🎤 Voice Cloning — Drop in any 5–30 second audio clip and it instantly learns the voice. Accent, tone, cadence — all of it.
🌍 31 Languages — English, Chinese, Japanese, French, Spanish, Arabic, Hindi, Korean and 23 more. Switch mid-sentence.
🎯 Pinyin / IPA / Phoneme control — Nail every pronunciation. Perfect for names, technical terms, multilingual scripts.
⏸️ Pause control — [pause 1.5s] anywhere in your text. Precise timing.
⚡ GPU Accelerated — NVIDIA RTX 2070+ with bfloat16 precision. Apple Silicon via Metal MPS.
🔒 100% Private — Everything runs on your machine. Your voice samples never leave your device.
| Platform | GPU | Status |
|---|---|---|
| Linux (Debian/Ubuntu) | NVIDIA RTX 2070+ | ✅ |
| Linux (Arch) | NVIDIA RTX 2070+ | ✅ |
| Linux (AppImage) | NVIDIA RTX 2070+ | ✅ |
| macOS (Apple Silicon) | M1/M2/M3/M4 | ✅ |
| Windows 10/11 | NVIDIA RTX 2070+ | ✅ |
Requirements: 8GB+ VRAM, 16GB system RAM, Python 3.10+
git clone https://github.com/xirtus/MOSSlanding
cd MOSSlanding
# Linux
cd release/linux && ./install.sh && mosslanding
# Build packages
cd release/packaging && ./build_all.sh
# Output in release/dist/release/
├── linux/ # Linux app source + install scripts
├── mac/ # macOS SwiftUI app + DMG builder
├── windows/ # Windows app + PowerShell installer
├── appimage/ # AppImage builder
├── packaging/ # Native package builders
│ ├── arch/ # PKGBUILD for Arch Linux / pacman
│ ├── debian/ # DEBIAN/ control files for .deb
│ ├── apt-repo/ # APT repository generator + one-liner installer
│ └── build_all.sh # Build everything in one shot
└── dist/ # Built packages output
This app is a native desktop launcher for the MOSS-TTS family by OpenMOSS — state-of-the-art open-source TTS that outperforms closed-source models like Doubao and Gemini 2.5 Pro in subjective evaluations.
Models used:
OpenMOSS-Team/MOSS-TTS-v1.5(~4-6 GB) — voice cloning & direct generationOpenMOSS-Team/MOSS-VoiceGenerator(~1-2 GB) — voice design from text descriptionsOpenMOSS-Team/MOSS-Audio-Tokenizer(~1 GB) — audio encoding/decoding
-
Push a version tag:
git tag v1.0.0 git push origin v1.0.0
-
GitHub Actions automatically builds
.deband.pkg.tar.zstand attaches them to the release. -
(Optional) Deploy APT repo to GitHub Pages:
cd release/packaging/apt-repo ./setup-repo.sh 1.0.0 git checkout gh-pages cp -r repo/* . git add . && git commit -m "apt repo v1.0.0" && git push
-
(Optional) Submit PKGBUILD to AUR for
yay -S mosslandingsupport.