PURPOSE BUILT AI ← all builds
voice-local v1 · windows · MIT + Apache licensed models

Broadcast voiceover that never leaves your machine.

Script in, narration out, entirely on your PC. Built on the open model independent blind tests preferred over ElevenLabs. Clone your own voice from a 10-second sample. No credits, no character meter, no $99/month.

⇩ Download free Setup + your brand voice — $97

Free forever. Readable Python you can audit yourself. Windows 10/11 + Python. GPU optional.

$ python voice_local.py course_intro.txt --voice me.wav
[synth] 14/14 chunks · cloned voice · loudness -16 LUFS
[done] course_intro_voiceover.wav · 3.2 min of narration · 100% on your machine

Why local beats the cloud voice tools

Cloud TTS ($22-99/mo)Voice Local
PricingCredits per character, retries burn credits, overages auto-billed$0, unlimited
Your scriptsProcessed on their serversNever leave your PC
Voice cloningPaid tiers, sample uploaded to their cloudFree, sample stays local
QualityExcellentOpen model preferred over ElevenLabs in blind tests
Works offlineNoYes, fully
Auditable codeNoYes, it's yours

Get narrating in minutes, or don't lift a finger

Free download

$0

The full tool, both engines. Unzip, pip install, narrate. README covers GPU setup, CPU mode, voice cloning, and mp3 output.

Download

Pro Setup

$97

I remote in, tune it to your hardware, clone your brand voice from a short sample, and narrate your first script with you before we hang up. 30 minutes.

Book it

Studio

$497

For teams that produce content: up to 10 machines, multiple brand voice presets, script-to-narration workflow training, 90 days of support.

Equip my studio

Questions

Is it really free?

Yes. No account, no credits, no character meter. The paid options are my time: white-glove setup with your brand voice ($97) or a studio-wide install ($497).

How good is the quality?

The default engine is Chatterbox, an MIT-licensed model that independent blind listening tests preferred over ElevenLabs. A lighter CPU engine (Kokoro, Apache license) is included for machines without a GPU.

Can it clone my voice?

Yes, from 7 to 20 seconds of clean speech, and the sample never leaves your machine. Clone only voices you own or have written permission to use.

Who built this?

Purpose Built AI — a systems engineer's practice that rebuilds overpriced cloud tools as local software and builds custom AI systems for businesses. This site runs on the same stack.

The Local Suite: four tools, one idea

Dictation, video clips, meeting notes, and voiceover. The subscription tools, rebuilt to run on your machine. All free, all readable code, all yours.

Flow Local Clip Local Meet Local Voice Local
Full Suite setup, all 4 tools, $197