Make the free voice from the blind test on your own computer: Python 3.13 setup for Mac and Windows, a complete copy-paste Kokoro script that downloads the model files and writes my-voice.wav, and a five-minute quality check for your real script

The companion to the reel. You heard two AI voices say the same sentence: A was ElevenLabs (Sarah, generated on paid account credits) and B was Kokoro v1.0 (preset af_heart, generated locally on a Mac with no subscription or API key). PAGE 1 — what was tested, what 'free' actually means here (you provide the computer, electricity and setup time; ElevenLabs also has a free tier), and what you need (Python 3.13, an internet connection for the one-time model download, roughly 337 MiB of disk). PAGE 2 — set up the workspace: the exact Mac Terminal and Windows PowerShell commands, no activation step needed. PAGE 3 — the complete self-contained quick_start.py: paste it, run it, it downloads the model and voice files on first run and writes my-voice.wav; edit one line to use your own text or another preset. PAGE 4 — test your real workload: the five-minute quality check (a 50 to 100 word paragraph, names, numbers, emotion, clipped endings), common fixes, and the official sources and licenses (MIT runtime, Apache 2.0 model). The optional starter ZIP (requirements, downloader, generator, quick start, README, no model weights) is linked below. Mac execution was tested; Windows commands are supplied but were not executed on a Windows machine. Independent guide from Hyperautomation Labs; not affiliated with Kokoro, ONNX, or ElevenLabs.

By submitting, you'll be subscribed to the Hyperautomation AI Report newsletter — three practical AI emails a week. Unsubscribe anytime.

Free. No spam. Unsubscribe anytime.