← Lab

LAB / 03 · Live tool

Local Blog Audio Generator

Generate natural-sounding article narration locally, privately and without a per-use fee.

Read the field note
Audio waveform and narration controls
03 / UTILITIES

01 / CONTEXT

The problem

Browser text-to-speech is convenient but inconsistent. Cloud narration can become costly and sends editorial content to a third party. The local workflow creates reusable audio files on the author's own machine.

02 / CAPABILITIES

What it does

  1. 01

    Turn queued article text into publishable MP3 files.

  2. 02

    Choose language-specific voices for English and Spanish.

  3. 03

    Move completed sources to a done folder and avoid duplicate generations.

03 / SYSTEM

Privacy & architecture

A local Python and FFmpeg pipeline triggered through an npm command; generated MP3 files become static site assets.

Built with

  • Python
  • MLX Audio
  • Kokoro
  • FFmpeg

04 / HONEST NOTES

Known limits

  • Generation speed depends on the computer and article length.
  • Pronunciation and pacing still need a listening pass before publishing.