Start with a script that can be changed

A voiceover often needs one small correction after you hear it: a name is pronounced oddly, a sentence runs too long, or the pacing feels wrong. Keeping each sentence as its own clip makes that correction local. You change that line and leave the rest of the recording alone.

Paste your whole draft first. TTS Lines separates sentences and paragraphs into editable lines. Read through the line list before generating. Join short fragments that belong together, and split lines that need a distinct pause.

Generate, listen, then adjust

Choose a main voice that fits the language of the script. You can give individual lines a different speaker. The local model has to load before generation begins; on first use, that can mean a large download. Once ready, make the voiceover and listen to each line, especially names, numbers, and abbreviations.

If a line needs work, edit its text, change its voice or speed, and generate only that line again. Move lines up or down if the order changes. A generated clip becomes stale after you change it, so make it again before export.

Choose the export that fits the edit

Use individual MP3 downloads when you are cutting a video or placing lines on a timeline. Use the ZIP export to take every clip at once. If you want a single continuous narration, export a joined MP3 or WAV. The SRT file uses the actual generated clip durations so captions line up with this export.

Listen to the final joined file before publishing. A natural pause in writing does not always translate into the same pause in speech, and proper nouns may still need phonetic spelling.