How to run XTTS-v2 locally

Coqui · Voice cloning · 460M params · Coqui Public Model License (CPML, non-commercial) (non-commercial)

The widely-adopted multilingual voice-cloning model that copies a speaker from ~6 seconds of audio across 17 languages.

What it costs to run — $0 markup

On your own machine — $0. Runs free if you have about 6.0 GB of VRAM (~4-6GB; clones a voice from a ~6s clip), via Coqui TTS.
Rent a GPU — from $0.26/hr. Fits RTX 3090 24GB at the direct vendor price ($0 markup) — pay only for the minutes you generate.
Download the weights — free. Open weights at coqui/XTTS-v2.

Note: generative-media models are billed per image / per second / per minute on hosted services — not per token. Running locally or on your own rented GPU is usually far cheaper and keeps your data on your machine.

Key facts

Does	Voice cloning
VRAM to run	~6.0 GB (~4-6GB; clones a voice from a ~6s clip)
Download	~1.9 GB
Parameters	460M
License	Coqui Public Model License (CPML, non-commercial) (non-commercial)
Run with	Coqui TTS

Get XTTS-v2 on Hugging Face →

More voice & audio models

Whisper large-v3 — Speech → text, ~10 GB VRAM
Whisper large-v3-turbo — Speech → text, ~6.0 GB VRAM
Kokoro-82M — Text → speech, ~1.0 GB VRAM
MusicGen Large — Text → music, ~16 GB VRAM
Chatterbox — Text → speech, ~8.0 GB VRAM
Parakeet TDT 0.6B v2 — Speech → text, ~4.0 GB VRAM

Browse: all media models · chat / LLM models