Sound effects and voices

For music, see Music.

Add effects from the library

1. In the Library, open Audio and choose Sound effects.
2. Type in the search box to narrow the list, and hover over a sound to hear it.
3. Click a sound to add it at the playhead, or drag it to a sound lane.

The sound effects come in the same free library download as the music. If the panel offers Download the library, click it once.

Have sounds made to measure

Zornov can make a sound for an exact moment, with no file needed. Ask the AI editor for it in plain words, and say when it happens:

• "A whoosh for the transition at 12 seconds, as long as the move."
• "A riser that peaks exactly on the drop at 30 seconds, then a boom."
• "A match strike when she lights the cigarette, distant period traffic under the whole street scene."
• "A café murmur under this scene, a door bell at 8 seconds."

The sounds are made on your computer by physical models of how things sound. Each new request makes a different take, so repeated effects don't sound copied. A sound can be placed to start at a time or to land its main moment (a riser's peak, a match catching) on a time.

What can be made:

• Transitions and hits: whooshes, risers, booms, hits, interface clicks and pops.
• Places (beds up to two minutes): stream, waterfall, rain, wind, birds, crickets, fire, waves, forest, night.
• Foley and spot effects: impacts, knocks, scrapes, squeaks and creaks, match strikes, lighters, a flame igniting, fire crackle, pouring, dripping, splashing, bubbling, engines, a car passing, traffic, horns, hooves, crowd murmur, a café, applause, clocks, switches, locks, doors, typing and typewriters, machine hum, servos, camera shutters, phones, heartbeat, breathing, thunder, explosions, gunshots, bells, sirens, sci-fi zaps, glass clinks and breaks, bouncing balls, pops (a balloon, a cork, bubble wrap), paper, cloth and sizzling.
• Animals: pig, horse, hen, sheep, dog, cat, cow, duck, rooster, crow, owl, frog, lion, elephant, donkey and mouse. Ask for a call ("dog growling", "cat purring") and a size.

You can shape a sound with words such as distance, size, speed, count, material (wood, metal, glass, stone, plastic, ceramic) and period ("1920s traffic" adds bulb horns and hooves).

Use effects with restraint

• Three to five effect cues in a short video is plenty. A whoosh on every move gets tiring.
• Start a whoosh with the move and make it as long as the move. Put an impact on the frame of contact.
• Keep a place's bed about 20 dB under the voice, and change it when the place changes.
• Half a second of silence before a big moment makes it hit harder than any effect.

Lay footsteps

Footsteps take a surface and a shoe:

Ask for them in words ("footsteps on wet stone in heels, walking away") with a count or a length, and a speed. Each step sounds different from the last.

For an animated character, ask the AI editor to lay its footsteps. They land on the frames where its feet touch the ground, on the surface of the place it stands, and sized to the character. Laying them again replaces the old ones.

Narrate with a voice

Narration is spoken by the voice service you choose in Settings > AI > Voice & sound, under Spoken by:

Mac voices is the choice when no voice service is set up. For the most natural sound, add Premium or Enhanced voices in macOS: System Settings > Accessibility > Spoken Content. Zornov lists those first.

To set up a service:

1. Open Settings > AI > Voice & sound.
2. Under Spoken by, choose the service.
3. Enter a Voice and, if you want, a Model.
4. Paste your key under API key and click Save. Zornov keeps it in your login Keychain.

If the chosen service can't speak a line (no key, no credit, offline), a Mac voice says it instead and Zornov tells you so.

To keep the built-in voices out completely, turn on Never use Mac voices. Then narration comes only from the chosen service, or not at all, for the AI editor and for any AI assistant you connect. A line the service can't speak fails and says why.

Make narration

Ask the AI editor to say a line: "Narrate at 4 seconds: Every map has an edge." You can name a voice of your service and a speed. The line goes on the Voice lane, levelled for speech, and any music under it dips while it plays. The AI editor also knows when each word is spoken, so it can time captions and pictures to the words, and a character's mouth can follow the line (see Characters and looks).

Sung parts

The AI editor can also sing a short part in soprano, alto, tenor or bass voices, on exact notes and in time. It sings syllables (doo, la, bap), not words, and the voice sounds plainly synthetic: good for a jingle hook or a cartoon choir, not a lead vocal.

Dub into another language

Dubbing speaks every subtitle of a translated track in each character's voice, fitted to when the original actor speaks. You do it in the Sound panel (⇧ X on Mac, Shift X on Windows and Linux), on the Subtitles & dubs tab.

1. Make subtitles of the original dialogue: click From dialogue, or Import… an SRT or VTT file. See Captions and subtitles.
2. Make sure each subtitle has its speaker. The AI editor can mark who speaks each line for you.
3. Translate the track: type the language code (for example fr) in the box beside Translate and click it. This uses your OpenAI key, so it works only on a Mac. You can also ask the AI editor to translate the subtitles.
4. Select the translated track. Under Voices in that language, type a voice for each character: a voice from the service chosen in Settings.
5. Click Dub.

Each character keeps one voice in that language for the whole film. The dubbed lines go on their own lane, and the language gets its own mix (music and effects with the dub) and its own audio track when you export. Zornov lists any line that won't fit even when spoken 25 % faster: shorten its translation and dub again.

Related pages

• Audio and mixing
• Music
• Captions and subtitles
• Animation and puppets