Skip to content

Voicebox Review

Voicebox is a free, open-source (MIT) desktop voice studio for Mac and Windows: clone a voice from seconds of audio, generate speech with several local TTS engines, dictate anywhere, and give AI agents a voice over MCP.

Free plan · from $12/yr · API

Stork's verdict on Voicebox

Voicebox is a generous free replacement for paid cloning and dictation apps if you have the disk and GPU for local models, with Mac users getting the most complete version.

Voicebox reviewed by Stork AI · stork.ai/en/voicebox

stork
I am one still image.
One typed sentence. This clip cost 5 cents, the minimum per video.
Make a photo talk for 20 cents a minute.
Upload one photo and type the words. No subscription. One free video a day.
Make one freeDevelopers: there is an API

What Voicebox does

Voicebox is a local-first AI voice app that pitches itself as a free, open-source alternative to ElevenLabs (speech output) and Wispr Flow (dictation) in one. You clone a voice from about 3 seconds of audio (an uploaded WAV, MP3, FLAC or WebM file, a mic recording of up to 30 seconds, or captured system audio), then generate speech of up to 50,000 characters per run; long text is split at sentence boundaries and crossfaded. You can switch engine per generation: the site lists Qwen3-TTS, Qwen CustomVoice, Chatterbox, Chatterbox Turbo, LuxTTS, TADA (Hume AI) and Kokoro, and the README adds Qwen VoiceDesign. Language coverage depends on the engine, from 10 languages up to 23 with Chatterbox. A Stories timeline editor handles multi-voice narratives, and an effects pipeline adds pitch shift, reverb, delay and compression.

For input, hold a global hotkey, speak and release: Whisper (99 languages) transcribes and, on macOS, pastes into the focused field, with optional clean-up of filler words by a local Qwen3 model. For agents, any MCP client (the site names Claude Code, Cursor and Cline) can call a voicebox.speak tool on 127.0.0.1:17493, and a local REST API exposes generation, profiles and history. Downloads are a DMG for Apple Silicon and Intel Macs and an MSI for Windows; Linux users build from source or use Docker. Acceleration covers Metal/MLX on Apple Silicon, CUDA, ROCm, Intel Arc and DirectML, or a remote machine.

Voicebox pricing

The desktop app is free and MIT-licensed. Paid cloud plans are announced but not yet available:

PlanPriceStatusIncluded
Local$0Available nowVoice cloning on every engine, dictation and capture, MCP/agent integration, unlimited local generations; no account
Cloud$12/year (launch price)Coming soonEverything in Local, plus end-to-end encrypted backup, sync across desktop and mobile, 25 GB storage, up to 5 devices, 30-day version history
Studio$48/yearComing soonEverything in Cloud, plus 250 GB storage, unlimited devices, 1-year version history, priority support

Voicebox says cloud pricing and limits are not final and will be confirmed at launch, and labels Studio pricing "TBD". There are no API keys, rate limits or per-character fees for local use. Your real cost is hardware and disk: models download separately, and a third-party review measured roughly 15 to 20 GB to install every engine.

Prices checked on the vendor's pricing page on November 10, 2026. See Voicebox pricing.

Key features

What Voicebox lists on its site and README:

  • Voice cloning from about 3 seconds of audio, from files, mic or system audio
  • Several switchable local TTS engines, including Qwen3-TTS, Chatterbox, TADA and Kokoro
  • Up to 50,000 characters per generation with automatic chunking
  • Global push-to-talk dictation with Whisper (99 languages) and optional local clean-up
  • MCP voicebox.speak tool and local REST API for agents and scripts
  • Stories timeline editor and an audio effects pipeline with presets
  • Native Tauri (Rust) desktop app for macOS and Windows; Linux from source or Docker

Who Voicebox is best for

Voicebox suits creators and developers who want voice cloning, TTS and dictation without a subscription or sending audio to the cloud, especially Mac users (where auto-paste dictation works) and anyone wiring a voice into Claude Code or Cursor via MCP.

If you need a hosted API, streaming TTS or broad commercial voice libraries, ElevenLabs is the paid standard. For dictation alone, Wispr Flow and Superwhisper are more focused. Windows and Linux users should note that dictation auto-paste is macOS-only for now.

What users say

A review of v0.5.0 by Andrew Ooo calls Voicebox a strong free local alternative to ElevenLabs and Wispr Flow, praising the engine choice, cloning, global dictation and built-in MCP server in a roughly 80 MB app. He notes heavy resource needs per engine (about 1 GB of VRAM for LuxTTS up to around 10 GB for TADA 3B on his Mac), no streaming TTS yet, no prebuilt Linux binaries, macOS-only auto-paste, and bug reports routed through Discord rather than GitHub issues. The GitHub repository shows about 60,000 stars. We found no independent side-by-side test of clone quality.

Pros

  • Free and MIT-licensed; no account, keys or per-character fees
  • Cloning, TTS and dictation in one desktop app
  • Choice of several open TTS engines per generation
  • Built-in MCP server gives coding agents a voice
  • Audio stays on your machine

Cons

  • Model downloads are large; all engines take roughly 15-20 GB
  • No prebuilt Linux binary; dictation auto-paste is macOS-only
  • No streaming TTS yet, per a v0.5.0 review
  • Cloud backup and sync plans are not launched

Frequently asked questions

Is Voicebox free?

Yes. The Local plan is free with unlimited local generations and no account. Paid Cloud ($12/year launch price) and Studio ($48/year) plans for backup and sync are marked coming soon.

Is Voicebox an ElevenLabs alternative?

It positions itself as one: voice cloning and TTS run on your own machine with open models, with no per-character fees. It does not offer ElevenLabs' hosted API or voice marketplace.

Does Voicebox run on Windows and Linux?

Windows has an MSI installer with CUDA or DirectML acceleration. Linux has no prebuilt binary yet; you build from source or run it in Docker.

How much audio does Voicebox need to clone a voice?

The site says about 3 seconds. Samples can be uploaded files (WAV, MP3, FLAC, WebM), a mic recording of up to 30 seconds, or captured system audio.

What languages does Voicebox support?

Speech generation covers 10 to 23 languages depending on the engine (23 with Chatterbox). Dictation uses Whisper, which covers 99 languages.

Can AI agents use Voicebox?

Yes. It runs an MCP server at http://127.0.0.1:17493/mcp with a voicebox.speak tool; the site names Claude Code, Cursor and Cline. A POST /speak endpoint serves non-MCP tools.

Compare Alternatives

See every Voicebox alternative, compared

Voicebox

Free plan · from $12/yr · API

From the Stork blog

Enjoying this? Get one like it in your inbox each morning.

one email a day · unsubscribe in two clicks · no third-party tracking