Introduction

One of the biggest things that sets VEO 3 apart from other text-to-video models is native audio. It can generate dialogue, ambient sound, and music — all from your prompt, in a single pass.

But to actually get good audio, your prompt needs a slightly different structure. You're not just describing a scene anymore; you're directing an actor.

In this article, we'll cover the audio + dialogue prompt formula I use in VEO 3, with examples you can copy.

If you haven't already, read the text-to-video prompt formula first — this article builds directly on it.

Quick Heads-Up Before We Start

On Veogen, audio is on by default in every mode — text-to-video and frames-to-video both. If you specifically want a silent clip (e.g. for adding your own soundtrack later), toggle audio off in the generation settings.

For lip-synced dialogue, text-to-video tends to be the most reliable starting point because the model has full control over the character's mouth shapes from frame one. Frames-to-video can still talk, but lip-sync precision depends on how clearly the reference image shows the face.

The Formula

Here's the structure for any scene where someone is speaking:

[CHARACTER] + [ACTION] + [CAMERA ANGLE] + [ENVIRONMENT/LIGHTING] +
[DIALOGUE] + [TONE OF VOICE] + [AMBIENT SOUND or MUSIC]

Or, written more naturally:

[CHARACTER] does [ACTION] in [ENVIRONMENT]. [Camera direction]. [Character]
says in a [tone of voice]: [dialogue]. [Ambient sound description].

It's the same 8-part formula from before, with two new pieces bolted on: dialogue and tone of voice.

1. Dialogue

This is the actual line(s) you want the character to say. VEO 3 will lip-sync the words to the character's mouth.

Examples:

• Today… we take back the throne.
• I haven't seen a sunrise like this in 30 years.
• Welcome back to Yeti Adventures — today, we're ice fishing.
• Are the rumors about war in the Outer Rim true?

⚠️ Avoid quotation marks around the dialogue. Quotation marks sometimes trigger VEO 3 to add subtitles to the video. Just write the line as plain text inside your prompt.

Keep dialogue short — usually one or two sentences. VEO 3 only generates 8-second clips, so anything longer than that will get cut off.

2. Tone of Voice

This is the how. Tone is honestly more important than the actual words, because it controls the entire performance.

What to specify:

• Emotion (nervous, angry, excited, sarcastic)
• Accent (British, southern American, Scottish, Japanese, etc.)
• Energy level (whispering, shouting, deadpan)
• Age and gender (young man, elderly woman, child)

Examples:

• Nervous young man stammering his words
• Confident commander shouting with intensity
• Elderly woman whispering in a tired voice
• Sarcastic tone with casual delivery
• Deep, gravelly Scottish male voice

The richer your tone description, the more "alive" the performance feels.

3. Ambient Sound and Background Music

This is what gives the world its texture. Even one ambient sound can completely change how a scene feels.

Ambient sound examples:

• Wind howling through the canyon
• Rain pelting the rooftop
• Birds chirping in a quiet forest
• Distant traffic and car horns
• Marketplace chatter

Background music examples:

• Soft piano playing in the background
• Lo-fi beats with a warm tape hiss
• Epic orchestral music swelling slowly
• Eerie low drones underneath everything

Specific sound events:

• A sudden thunderclap
• Glass shattering
• Boots crunching on gravel
• A door creaking open

💡 Don't overload your prompt. Pick 1–2 ambient sounds max. More than that and VEO 3 starts mixing them weirdly.

A Full Example, Broken Down

Let's piece it all together:

A young man in a futuristic flight suit stands on a landing platform at
sunset. The camera pushes in slowly from chest height. The scene is lit
by the warm glow of a setting alien sun. The video has a sci-fi drama
tone. He looks out at the sky and says, in a quiet but determined voice:
I don't know what's out there… but I'm ready to find out. Wind howls in
the background and low ambient synth music plays.

Walking through it:

• Character: young man in a futuristic flight suit
• Action: stands on a landing platform, looks out at the sky
• Camera: slow push-in from chest height
• Environment + lighting: landing platform at sunset, warm alien sun
• Style: sci-fi drama
• Dialogue: "I don't know what's out there… but I'm ready to find out."
• Tone: quiet but determined
• Ambient + music: wind howls, low ambient synth

That single prompt gives VEO 3 everything it needs to make a fully voiced, atmospheric, cinematic clip.

A Two-Character Dialogue Example

Multiple characters can talk in the same scene too, as long as you're clear about who's speaking.

A mother in a house robe pours tea in the kitchen and leans on the
counter. A tripod-style medium shot frames her. Warm interior kitchen
lights with morning sun peeking in. Style: family drama. The kettle
steams, birds chirp outside.

She says in a soft, concerned voice: You've barely eaten. Is everything
okay at school?

The camera cuts to a side-angle close-up of her teenage son at the
table, pushing eggs around with his fork. He replies in a tired,
dismissive tone: It's just exams. I'm fine.

Notice how each character has their own description, action, camera angle, and tone — that's the trick to making multi-character scenes feel real.

Common Mistakes With Audio Prompts

A few things that consistently break dialogue scenes:

• Using quotation marks — triggers subtitles
• Long monologues — anything over 2 sentences usually gets cut off
• Forgetting the tone — VEO 3 will pick a neutral, robotic delivery if you don't specify
• Stacking too many sounds — pick one or two, not five
• Forgetting to enable audio — if you turned audio off in settings, no amount of prompting will bring it back; double-check the toggle

Conclusion

Audio is what takes VEO 3 from "cool AI demo" to "this looks like a real movie clip". Once you nail the dialogue + tone formula, your videos start to feel alive — characters have personality, scenes have atmosphere, and your content stands out fast.

Next up, we'll cover one of the most underrated parts of cinematic prompting: camera angles, shots, and movements in VEO 3.