Prepare the text for speech
Write the way you want the sentence heard. Spell out unusual abbreviations and check names. Short lines are easier to revise, but do not split a natural phrase just to make every line the same length.
Paste the text into TTS Lines and read the line list before generating. Join a fragment that belongs to the previous sentence or split a line where the speaker changes.
Generate and listen
Choose the language and a local voice. The model downloads on first use, which can take longer than later visits. Generate the script and listen through it once. Change a weak line's wording or speed, then generate only that line again.
Pauses between lines are part of the full playback and export. Adjust them after hearing the joined take, especially at a paragraph break.
Choose the handoff
Export MP3 for an easy full recording. WAV gives you an uncompressed full file. A ZIP holds one MP3 per line when you need to move narration around a video timeline. The SRT file follows the generated line durations, so make it from the final take.