blog-audio

Blog Audio is a skill in the claude-blog toolkit that turns written blog posts into spoken audio using Google's Gemini text-to-speech. It solves the problem of making written content accessible in an audio format, letting authors offer a listen-along option or a podcast-style version of an article without leaving their writing workflow. It can be invoked directly through the `/blog audio` command or triggered internally by the companion blog-write skill, and it degrades gracefully by returning silently when no API key is configured so it never blocks the writing process.

The skill offers three narration modes: a 200-300 word spoken summary, a full article read-aloud, and a two-speaker podcast/dialogue mode. It ships a catalog of 30 prebuilt Gemini voices (each with a tonal characteristic such as Informative, Firm, or Breezy), supports 80+ languages via automatic detection, and recommends voice pairings by content type. Claude prepares the spoken text while a bundled Python script (invoked through a `run.py` wrapper that manages a local virtual environment) performs the actual TTS call, producing MP3 output with HTML5 embed code. It documents current Gemini TTS model IDs and uses FFmpeg for WAV-to-MP3 conversion, falling back to WAV when FFmpeg is missing.

The primary users are bloggers, content marketers, and publishers who want to add audio versions or podcast segments to their posts. Typical use cases include generating a quick audio overview of a long guide, producing a full narrated read of an article, or creating a conversational two-host podcast from written content, all with a chosen voice tuned to the blog's tone.