Chapter 3 · Produce
Give it a voice.
You already wrote the script — it’s the talking points on every scene. assemble --voice speaks them over the captured video: silence-padded, sync-safe, and rendered straight into the MP4.
Local-first, pluggable anywhere
No account, no API key required: the default provider is macOS say. Any TTS CLI drops in through a command template, and a custom engine is just a function — the config is real JavaScript.
voice: {
enabled: true,
provider: 'say', // macOS built-in
// command: 'espeak -f {textFile} -w {outFile}',
overflow: 'extend', // long narration freezes the last frame
}
That button is your browser’s speechSynthesis standing in — the real pipeline renders AAC into the MP4 with ffmpeg, and refuses to stitch if any part’s audio shape is wrong.