Use MP3 for a finished listen
The full MP3 joins every generated line with the pauses you set in the editor. It is the simplest file to send for review or place under a rough video cut. Listen through it once before sharing, especially where the speaker changes.
You can choose 128 or 192 kbps in TTS Lines. That setting changes the export size, not the voice model or the words it produces.
Use WAV when your editor wants uncompressed audio
WAV keeps the joined recording uncompressed. It is useful when you plan to process or mix the whole narration in an audio editor. The file is larger than MP3, so MP3 is often easier for a quick handoff.
If you need only one corrected sentence later, download that line as an MP3 instead of exporting the whole recording again.
Use ZIP when the timing may change
The ZIP contains one MP3 per line, named in order. Put the clips on separate spots in a timeline, move a callout, or leave room for music. This is the practical choice when picture changes after the script has been recorded.
The SRT option is a separate text file. Its cues use each generated clip's duration and the pauses set in TTS Lines. Moving clips in another editor means the captions need to move there too.