ferrants/skill-tts

{"i": 0, "repo": "ferrants/skill-tts", "stars": 1, "lang": "Python", "ok": true, "char_count": 1850, "text": "This is a plugin for Claude Code that converts written text into spoken audio using OpenAI's text-to-speech service. You can use it to narrate a document, create a voiceover for a video, or simply listen to something instead of reading it on screen.\n\nInstallation is a single command using the skills package manager. You can install it globally or scope it to a specific project. To use it, you need an OpenAI API key stored in a file in your home directory. The setup instructions include a clever trick to keep the key out of your shell history, which is a nice security touch.\nThe main skill, called tts, does more than just call the API. It tries to figure out your intent from how you phrase things. If you say \"read this to me,\" it plays the audio through your speakers. If you say \"generate an mp3,\" it saves a file without playing anything. It also rewrites text to sound natural when spoken aloud, converting tables, bullet points, and code blocks into something a person would actually say. You can ask it to read verbatim when the exact wording matters, like with ad copy or legal text.\nLong texts are handled automatically. The OpenAI API limits input to about 4,000 characters, so the script breaks longer content into chunks, processes them in parallel, and stitches the audio back together. A 6,000-word article takes seconds. The script also works around a quirk where the newer voice model ignores speed settings, fixing the tempo locally while preserving pitch.\nThere are a few caveats worth noting. Your text is sent to OpenAI's servers for processing, so you should be careful with confidential documents or anything under a nondisclosure agreement. The audio file itself is created and stored on your own machine. Also, OpenAI requires you to disclose that a voice is AI-generated if you publish the audio, which matters for ads, videos, or podcasts but not for personal use.\nThe tool requires Python 3, though it uses only the standard library so there is nothing extra to install. A program called ffmpeg is needed for longer texts and speed adjustments. A media player is optional if you want to play audio directly from the command line.", "error": null}