Most people watch your film muted. Choose its sound anyway

A product film has two audiences, and they get two different films. The first, nearly everyone, sees it autoplay in a feed or a gallery with the sound off. The second taps, or pressed play on purpose, and hears whatever the file carries. You make the film for the first audience. You choose its sound for the second.

Nitaris asks the question before you generate, next to the length and the shape, and the answer is frozen onto that film. It is worth thirty seconds of thought, because the default is not silence.

Sell it silent first

Before choosing anything, watch a render muted, start to finish. If the product and the point of the film land without sound, it is ready for the feed. If the film only makes sense once you hear it, because a voice carries the message over footage that never shows it, it will fail for most of the people who see it, and no choice of sound fixes that.

Three choices, and what each one is for

• Natural, the default. The ambience and foley the model rendered with the shot: the room tone of the set, a cap clicking shut, a pour, fabric moving. It is texture, not a soundtrack. It suits a product page video someone presses play on, where total silence can feel like a fault.
• Silent. An MP4 with no audio track at all. Pick it when you are going to add music in the app you post from, so there is nothing of ours to strip out first, or when the film will only ever run muted.
• Voiceover. One spoken line over the film's own sound, which is turned down underneath it. You can switch that bed off and speak over silence. Pick it when a single sentence says something the picture cannot.

When a voiceover earns its place

A voiceover is worth it when the most persuasive fact about your product is invisible. What a serum contains. That a bottle is refillable. That a strap is vegan leather. A camera cannot show any of those, and a caption only holds a few words before it starts covering the product.

It is not worth it when the line describes what the viewer can already see. "A beautiful gold pendant on a dark background" spends the one sentence the film has room for on information the first frame already gave away.

Writing the line

• Fit the length. The studio shows how many words fit: about 26 in ten seconds. Treat that as a ceiling, not a target. One sentence well under it leaves room for the picture.
• Say the product's name once, early, the way a customer would say it out loud.
• One claim per line. The second claim is the one nobody remembers.
• Do not open with a greeting. The voice already starts a beat after the first frame so the picture lands first; a "Hi there" wastes that beat.
• If you are unsure, leave the script empty. Nitaris writes a line from the shot, and a draft to react to is faster than a blank box, even if you rewrite every word.

Pick the voice for the look, not for variety

Six voices are offered, and they differ in register more than in anything else: an even narrator, which is the default, an upbeat quick read, a low and measured one, a bright excitable one, a soft close one, and a clear editorial one. Match it to the mood you chose for the film. A low, measured read over a moody luxury treatment agrees with the picture. The same voice over a bright pop-art film argues with it.

Write the words down as well

Whatever the voice says, put it in the listing's description or the post's caption too. Most viewers will never hear it, and a shopper who cannot hear at all never will. An earlier post, "alt text is the photo, in words", makes the same argument for your stills: anything the medium carries has to exist somewhere as text.

Music is a separate decision

Nitaris does not add a music track. If you want one, choose Silent before you generate, then add licensed music in the editor you publish from. The answer on sound covers why that decision should stay with you.

The short version

Watch it muted first. Keep Natural for a product page, choose Silent when you will add music, and use a voiceover only for the one fact the picture cannot show. Keep the line short, say the name once, match the voice to the look, and write the line down beside the film.