AI Video Prompts for Street Interview Videos

The AI street interview was the first format that made people realize AI video could talk. When Google’s Veo 3 arrived with native sound in 2025, feeds filled with man-on-the-street clips where the passerby, the interviewer, the traffic and every word were generated. Many viewers could not tell. The format still works in 2026 because the structure is so familiar: a question, a beat of hesitation, an answer nobody expected.

Unlike most camera-style prompts, a street interview lives or dies on dialogue timing. The visuals are simple (two people on a sidewalk, a mic, a busy background). The hard part is fitting a question and a punchline into a few seconds without the characters speed-talking, and getting the audio mix to sound like a real handheld recording instead of a studio.

Below are prompts for the classic vox pop, multi-speaker exchanges, non-English interviews, fictional characters and series formats. For general dialogue technique, the Veo 3 prompts guide is a good companion.

Why Street Interviews Need Their Own Prompt Approach

Four things separate a convincing AI vox pop from a stiff one:

• Framing that feels grabbed. Vertical 9:16, handheld at eye level, the interviewee framed chest-up and slightly off-center, the mic poking into frame from one side. Pedestrians walk behind them, sometimes glancing at the camera.
• A visible, cheap mic. A handheld foam-covered mic held by the interviewer, or a small wireless clip mic on the collar. Describe it, but skip brand logos and news outlet names.
• Dialogue formatting. Speaker description, colon, line in quotes. This pattern is widely recommended for Veo because it tells the model exactly who speaks and what they say.
• Real street audio. Traffic, a bus pulling away, distant chatter, wind on the mic. The voices should sit slightly above the noise, not in a silent bubble.

The Street Interview Prompt Formula

Vertical handheld street interview video, eye level, [city street + time of day + weather].
[Interviewer description] holds a [handheld foam mic / clip mic] toward [interviewee description], framed chest-up, pedestrians passing behind.
Interviewer asks [tone]: "[short question]"
[Interviewee] [reaction: pauses / squints / laughs], then answers [tone]: "[short answer]"
Audio: city ambience, [specific street sounds], voices slightly above the noise.
Natural daylight, no color grading, (no subtitles). [8 seconds].

Ready-to-Use Street Interview Prompt Templates

1. The Classic Hot Take

One question, one confident old man, one punchline.

Vertical handheld street interview video, eye level, busy downtown
sidewalk on a sunny afternoon. A young female interviewer in a denim
jacket holds a black foam-covered mic toward a man in his seventies
with a flat cap and grey beard, framed chest-up, pedestrians walking
past behind him. The interviewer asks brightly: "What's the most
overrated thing in life?" He thinks for a second, then says flatly:
"Brunch. It's just late breakfast." Audio: traffic, distant chatter,
footsteps. Natural light, no color grading. (no subtitles) 8 seconds.

Why it works: The pause before the answer is written into the prompt, and both lines are short enough to be spoken at a natural pace inside 8 seconds.

2. The Price Guess

A proven vox pop game with an instant reaction.

Vertical handheld street interview video outside a coffee shop on a
cool morning. An interviewer in a green beanie holds a small paper cup
of coffee and a foam mic toward a woman in her thirties in a camel coat.
The interviewer asks: "How much do you think this coffee cost?" She
says confidently: "Four dollars?" The interviewer shakes his head. She
covers her mouth and laughs: "No. Stop." Audio: espresso machine through
the open door, traffic, wind on the mic. (no subtitles) 8 seconds.

3. Two Friends, One Interrupts

Multi-speaker exchanges feel more spontaneous.

Vertical handheld street interview video on a university campus path,
late afternoon. An interviewer holds a mic toward two friends in
hoodies, one tall with curly hair, one short with glasses. The
interviewer asks: "Who's the better cook?" The tall one starts: "Oh,
definitely me, I" and the short one cuts in, pointing: "He burned
cereal." Both burst out laughing. Audio: students chatting, a bike
bell, wind. Handheld, slight shake. (no subtitles) 8 seconds.

Why it works: The cut-off line and overlapping laugh are what real interviews sound like. Veo 3.1 supports multiple speakers in one clip.

4. Time Traveler on the High Street

Clearly fictional characters are the safest way to go viral with this format.

Vertical handheld street interview video on a crowded shopping street,
overcast day. An interviewer in a yellow raincoat holds a mic toward a
Roman legionary in full armor and red cloak, looking confused. The
interviewer asks: "What do you think of modern life?" The legionary
points at a passing electric scooter and says gravely: "Your chariots
are too quiet. It is unsettling." Shoppers stare as they pass. Audio:
city buzz, scooter whir. (no subtitles) 8 seconds.

5. The Dog Has Opinions

Talking animals are obviously AI, which makes them easy to post honestly.

Vertical handheld street interview video in a sunny park. An
interviewer crouches and holds a mic down toward a golden retriever
sitting next to its owner's legs. The interviewer asks: "What's the
worst word in English?" The dog looks up, mouth moving naturally, and
answers in a deep calm voice: "Bath." The owner laughs off camera.
Audio: birds, kids playing in the distance, wind on the mic. Eye level
with the dog. (no subtitles) 7 seconds.

6. Rainy Night Under Neon

Weather and lighting change the mood of the whole series.

Vertical handheld street interview video at night in a busy city street
in light rain, neon shop signs reflecting on wet pavement. The
interviewer holds a clear umbrella over a young man in a black puffer
jacket and a foam mic toward him. The interviewer asks: "What's keeping
you up at night?" He grins: "Honestly? This umbrella's leaking on me."
Raindrops on the lens edge, slight noise in the shadows. Audio: rain,
tires on wet road, distant music. (no subtitles) 8 seconds.

7. Spanish-Language Vox Pop

For non-English channels. Kling 3.0 and Seedance 2.0 both lip-sync Spanish.

Vertical handheld street interview video on a sunny plaza in Madrid,
cafe tables in the background. An interviewer holds a mic toward a
cheerful grandmother in a floral blouse with a shopping bag. The
interviewer asks in Spanish: "¿Cuál es el secreto de una vida feliz?"
She laughs and answers in Spanish: "Siesta todos los días, y no
discutir." Audio: plaza chatter, church bells in the distance, cups
clinking. (no subtitles) 8 seconds.

8. Series Clip: Same Question, New Stranger

Write one template and swap only the interviewee to build a compilation.

Vertical handheld street interview video, eye level, same busy crosswalk
corner in daylight. Interviewer: a woman in her twenties with a black
bob haircut, red bomber jacket, holding a white foam mic. She holds the
mic toward [INTERVIEWEE: e.g. a construction worker in a hi-vis vest]
and asks: "What's a skill everyone should learn?" [He/She] answers
[TONE]: "[ANSWER, under ten words]" Audio: traffic, crosswalk beeping,
wind on the mic. (no subtitles) 8 seconds.

Why it works: The interviewer, location and question stay identical, so the cut between clips feels like one real afternoon of filming.

Model Tips for AI Street Interviews

Veo 3.1 is the successor to the model that started this format. Its native audio handles dialogue, sound effects and ambience together, and it supports native vertical output. Clips run up to 8 seconds, so budget your words. The Veo 3 prompt generator page covers how to structure speech.

Kling 3.0 generates lip-synced dialogue in English, Chinese, Japanese, Korean and Spanish, with accent control for American, British and Indian English. Single shots run up to 15 seconds, which gives longer answers room. See the Kling AI prompt generator.

Seedance 2.0 lip-syncs dialogue in eight or more languages and takes image references, useful for keeping your interviewer consistent across a series. Guides for the model suggest keeping lines to about five to ten words. More in the Seedance 2.0 prompt guide.

Hailuo 3.0 outputs native stereo audio in clips up to 15 seconds. Test short lines first and check lip sync before scaling up.

Common Street Interview Prompt Mistakes

• Long answers. The number one cause of chipmunk-speed speech. One sentence per speaker.
• Forgetting “(no subtitles)”. Models often burn in captions with misspellings. Add your own captions in the editor.
• Studio-clean audio. Real vox pops have wind and traffic. Silence behind the voices sounds dubbed.
• Real logos and outlets. Do not put news channel names or brand logos on the mic. It implies a real broadcaster.
• Lip sync over wide shots. Keep faces chest-up. The further away the speaker, the worse the sync looks. See talking-head prompt tips.

Label It as AI

The AI street interview is persuasive because it looks like public opinion. Do not generate interviews with real people, celebrities or politicians, do not present made-up answers as what “people in [real city] think”, and do not use generated passersby as customer testimonials. Label the video as AI in the caption and with the platform’s disclosure tool. Fictional characters and obvious absurdity are your friends here.

How LzyPrompt Generates These

Dialogue prompts are where formatting mistakes cost the most credits. In LzyPrompt, you type a premise like “street interview asking grandmas their dating advice”, and get 10 variations that each vary the interviewee, the city, the weather and the punchline, each written as a full scene prompt covering the exchange, the mic and framing, and the city ambience in the sound cues. Each one comes with a caption for posting. Pick a Camera Style such as handheld phone video, or leave it on auto-detect, and export the batch to CSV for a full compilation.

Try LzyPrompt free for 7 days and write a week of vox pops in one sitting.