Voice cloning

A guide to the best open-weight text-to-speech models you can self-host in 2026, ranked by Artificial Analysis Speech Arena Elo. Compare Step Audio EditX, Fish Audio S2 Pro, Voxtral TTS, Kokoro 82M, Maya1, NVIDIA Magpie, Chatterbox, and Zonos on quality, licensing, hardware, and deployment.

Voicebox is an open-source, local-first AI voice studio for cloning voices, dictating text, and composing multi-track audio. This guide shows how to run it as a server and expose it remotely with Pinggy.

VoxCPM2 is a 2B-parameter open-source TTS model with voice cloning, voice design from a text description, and a drop-in OpenAI-compatible API. Here's how to run it on your local GPU and expose it publicly with Pinggy.