TTS-WebUI

AI & Audio

One local web interface for many text-to-speech and voice models.

TTS-WebUI is one web interface for many open text-to-speech and audio models. You can pick Kokoro for fast, light English speech, Chatterbox or XTTS for voice cloning, Bark for expressive audio and more, then generate speech from your own text entirely on your own GPU. It has a simple React interface and a larger Gradio interface with extensions, a pipeline builder, speech-to-text, audio and music tools and an outputs browser. Everything runs locally, so your text and any reference voices stay on your network. Use voice cloning only with your own voice or with clear permission.

License

MIT

GitHub stars

3,292

Official website

https://github.com/rsxdalv/TTS-WebUI

Source code

https://github.com/rsxdalv/TTS-WebUI

Compatibility

Architectures

amd64

Database

Stateless

Resource footprint

4096 MB RAM · 4 CPU

Measured on 2026-10-22 (version v1.5.2 plus a few commits) in a Proxmox LXC (6 vCPU, 12 GB RAM) with an RTX 3090 passed through. The web UI idles at about 0.5 GB of RAM and about 2.2 GB after loading the Kokoro model. Generating an 8-second English sentence with Kokoro-82M took about 1.7 seconds and used about 1.2 GB of VRAM. Larger models (Bark, XTTS, Chatterbox) need more VRAM and RAM. The installation with its extensions is large: about 22 GB for the project folder in our LXC, so plan for 60 GB of disk or more. The first Kokoro run also downloaded a small language package automatically.

Proof of Work Gallery

Models
Generate speech
Resource usage

Deployment

# TTS-WebUI is installed with its own one-click installer (see the repo README),
# not Docker. We run it natively in an LXC as a systemd service (our real unit).
# /etc/systemd/system/tts-webui.service
[Unit]
Description=TTS-WebUI
After=network-online.target
Wants=network-online.target

[Service]
Type=simple
WorkingDirectory=/root/TTS-WebUI
# A tmpfs /tmp is too small for model installs, so point TMPDIR at disk.
Environment=TMPDIR=/root/tmpdir
ExecStart=/root/TTS-WebUI/installer_files/mamba/micromamba run -p /root/TTS-WebUI/installer_files/env python server.py --docker
Restart=on-failure
RestartSec=10

[Install]
WantedBy=multi-user.target

Alternative to

ElevenLabs, Google Cloud Text-to-Speech, Amazon Polly