Fredo Bang “Oouuh” ukulele duet · generators compared

What You Want Me To Do AI Video: Which Generator Makes the Real Oouuh Duet

Short answer: pick a generator that follows the original take and keeps its sound. This one does that in your browser — two photos (or one photo twice), no CapCut template, no prompt, no watermark.

The “What You Want Me To Do” AI video is the Fredo Bang “Oouuh” ukulele duet trend: a 2018 kitchen clip with Einer Bankz strumming a ukulele on the left and Fredo Bang singing the hook on the right, remade with people’s own photos on TikTok since October 2026. The phrase is a line from the hook, not the song title — which is why searching it mostly turns up general AI video tools.

  • 1–2 photos · people or pets
  • 10 / 15 / 23 s · 9:16
  • Original take audio
  • Default 136 credits (Seedance 2.5, 23 s, 480p)

Photo 1 · Ukulele player (left)

One person or pet per photo. The same photo in both slots makes a solo duet.

Photo 2 · Singer (right)

One person or pet per photo. The same photo in both slots makes a solo duet.

0 of 2 photos ready

JPG, PNG or WebP, 300–6000 px per side, up to 30 MB. One person or pet per photo, face visible and well lit; a full-body photo keeps your own clothes. Only use photos of yourself, your pets or adults who agreed.

Background
The default background: an empty narrow kitchen at night

The default is an empty kitchen like the original. Upload a photo of another empty place — your kitchen, a living room, a stage — and the duet moves there.

Model
Resolution
Video length

480p · 9:16 · 23 s · MP4 with the original sound

This video costs 136 credits. Charged when you start, refunded automatically if it fails.

This model needs purchased credits. Sign-up credits work with Seedance 2.0 Mini and Fast.

What the duet looks like

Two friends in black tees

Two friends in a home kitchen; the left one strums, the right one sings the hook. 14 s with the original sound.

More examples

The name trips up search

Not every “AI video generator” makes this one

Type the phrase into a search engine and the top results are general AI video makers — Canva, Adobe Firefly, invideo, Picsart — that turn a prompt into a clip. Ask a chatbot and it often assumes you want a talking avatar or a stage performance. None of them knows the kitchen duet.

The trend is one specific video: Fredo Bang sings the hook of “Oouuh” while Einer Bankz strums beside him, filmed in a narrow kitchen in 2018. People remember the line “what you want me to do”, so that is what they search. These nearby searches are different things:

  • “AI that does what you want” or “AI that makes anything you want” — general prompt-to-video or assistant tools, not this trend.
  • “I want to make an AI video” or “AI video generator maker” — broad how-to searches; the duet needs a template that already knows the moves.
  • Older songs with similar titles, such as Jimmy Reed’s “Baby What You Want Me to Do” — unrelated to the AI trend.

So the useful question is not “which AI video generator is best” but “which one reproduces this clip with my faces”. The checklist below is how to tell.

The closest-to-the-original checklist

A convincing version gets all seven right. Use them to judge any tool, including this one.

  1. 1

    Both faces are the ones you chose

    Two photos, two people (or pets), each keeping their own face, hair and clothes — not a random singer.

  2. 2

    Ukulele on the left, singer on the right

    The player strums with the mouth closed the whole time; the singer does every move.

  3. 3

    The narrow kitchen

    A small home kitchen at night, a door in the back, shot straight on from chest height like a phone video — or a room you picked on purpose.

  4. 4

    The original moves and timing

    The chin-hand sway, the grin, the raised fist, and at about 14 s the head thrown back for the long “ooh”.

  5. 5

    The original sound

    The real ukulele and vocal of the 2018 take, not a newly generated tune or silence.

  6. 6

    A length that reaches the hook

    10 s stops before the big “ooh”, 15 s ends as the head goes back, 23 s has the hook twice.

  7. 7

    Vertical, clean, ready to post

    A 9:16 MP4 without a site watermark, at a resolution that holds up on TikTok.

Every route scored against the checklist

There are seven kinds of tool people try for this trend. Here is what each one keeps.

What each kind of tool keeps from the original duet
RouteYour two facesKitchen and movesOriginal soundEffortVerdict
This page’s generatorTemplate built from the original takeYes, 2 photos or 1 photo twice; pets tooYes: kitchen or your own room, moves from the 2018 takeYes, the take’s own audioUpload and clickClosest result without editing
Dedicated sites that follow the originale.g. OOUUH AI’s Trend mode, StarrdYes, usually 2 photosYesYes, they say they keep itUpload and clickAlso close; compare length and price
Dedicated sites that use a prompt onlye.g. Kavel, MICT, aifredobangvideo.proYes, 2 photosSimilar scene, but the model invents the movesNo: newly generated audioUpload and clickLooks like the trend, sounds different
CapCut or PixVerse templatesFound through TikTok tags like #pioneertemplateDepends on the templateDepends on the templateOften, or add the sound in TikTokFind the right template firstGood if you already have one
General prompt appsKling, Runway, VeoHard to keep two facesOnly what your prompt describesNoLong prompts and retriesFor your own variation, not a remake
Talking-avatar toolsHeyGen, Synthesia, D-IDOne faceNo: a head talking to cameraNo: lip-sync to a scriptScript and voiceA different kind of video
Free photo-animation appsHedra, PicAlive and similarOne photo comes aliveNo duet, no ukuleleNoEasyFree, but not this trend

Tools change their features often; this table describes the kind of tool, and the next one lists specifics.

Specs and price per video, side by side

Dedicated tools differ a lot in length, sound and price, and few show a per-video price. These are the figures each one published.

As listed on each tool’s own pages in October 2026 (last read 2026-10-23); check before buying
ToolLengthPhotosSoundWatermarkPrice per video
RoadsterVideo (this page)10, 15 or 23 s; 480p–1080p1–2, pets OK; optional backgroundOriginal takeNone15–532 credits ($0.75–$26.60 at the lowest credit price); one-time packs
OOUUH AI (oouuhai.com)15 s, 9:162 single portraits (Original Song mode: 1 group photo)Original soundFree 5 s preview carries a mark20 / 40 / 80 credits at 480p / 720p / 1080p; smallest pack $9.99 for 10 credits
Starrd (getstarrd.app)23 s, 480p1–2, pets OKOriginal audioNot stated150 Starrd credits
aifredobangvideo.pro8 s, 720pNot statedAI-generated audioFree preview discontinued$9.90 for 1, $19.90 for 3, $29.90 for 5
Kavel AI10 s, 9:162Its own ukulele groove and a wordless “ooh”, not the songNot stated360–2,365 Kavel credits
MICT10 s, 720p2 adult portraitsNewly generated, not the songNot stated300 MICT credits
CapCut / PixVerse templatesSet by the templateUsually 1–2Set by the templateDepends on the app and planDepends on the app and plan

Credits on other sites are not comparable one to one: each sells them at its own price. “Not stated” means the tool’s page did not say.

Which one should you pick?

You want the real duet with two people you choose and the original sound
Use a generator that follows the original take — this page or another dedicated site in the first rows above. Seedance 2.5 makes the whole 23 s.
You already found a CapCut or PixVerse template you like
Use it inside the app; check that it has two slots and the kitchen moves before you pay for a plan.
You want a different scene — a stage, a beach, your living room
Here, upload your own background and keep the same moves. For a completely new routine, a prompt app like Kling or Runway is the tool.
You only want a face talking or lip-syncing
That is a talking-avatar video; HeyGen-style tools make it. It will not look like this trend.
You want to try it once for free
Sign-up credits here cover a short Seedance 2.0 Mini version at 480p where you are eligible; free photo-animation apps cannot make the duet at all.
It is for a brand or an ad
The song is not licensed for ads. Make the video, then replace the audio with music you have the rights to and turn on the platform’s AI label.

What this generator does and doesn’t do

It does

  • Put Photo 1 on the ukulele (left) and Photo 2 on vocals (right); the same photo twice makes a solo duet.
  • Work with pets: they stand upright, strum with whole paws or sing with their own muzzle.
  • Use the empty kitchen, or a background image you upload.
  • Follow the moves and timing of the first 23 s of the 2018 take and lay its own audio under the video.
  • Offer Seedance 2.0 Mini, Fast and Pro, Seedance 2.5 and MiniMax H3, priced before you generate.
  • Deliver a 9:16 MP4 with no site watermark, and refund every credit if a video fails.

It doesn’t

  • Take a text prompt or change the routine: the moves are fixed.
  • Split a group photo into two people — crop each person into their own photo first.
  • Lip-sync to your own voice or another song.
  • Make landscape (16:9) videos or anything longer than 23 s.
  • Accept photos of children, celebrities or people who didn’t agree.

Credits per video on this page

One price per video, shown before you generate, the same for one or two photos and for any background; it includes the still of your duo. Dollar values use the lowest credit price on sale, the Studio pack ($99 for 1,980 credits, $0.05 a credit). No subscription.

Default136credits

Seedance 2.5 · 23 s · 480p · $6.80

Lowest15credits

Seedance 2.0 Mini · 10 s · 480p · $0.75

Free on sign-up3credits

Eligible new accounts get 3 credits; the cheapest duet costs 15.

All 25 model, length and resolution prices

Seedance 2.5 is the only model for 23 s. MiniMax H3 renders at 768P and is listed as 720p. Paid-only models need purchased credits.

Credits per finished video
ModelLengthResolutioncreditsValue at the lowest credit price
Seedance 2.0 Mini10 s480p15$0.75
Seedance 2.0 Mini15 s480p20$1.00
Seedance 2.0 Mini10 s720p27$1.35
Seedance 2.0 Mini15 s720p39$1.95
Seedance 2.0 Fast10 s480p27$1.35
Seedance 2.0 Fast15 s480p39$1.95
Seedance 2.0 Fast10 s720p55$2.75
Seedance 2.0 Fast15 s720p80$4.00
Seedance 2.0 ProPurchased credits only10 s480p43$2.15
Seedance 2.0 ProPurchased credits only15 s480p63$3.15
Seedance 2.0 ProPurchased credits only10 s720p89$4.45
Seedance 2.0 ProPurchased credits only15 s720p132$6.60
Seedance 2.0 ProPurchased credits only10 s1080p219$10.95
Seedance 2.0 ProPurchased credits only15 s1080p327$16.35
Seedance 2.5Purchased credits only10 s480p61$3.05
Seedance 2.5Purchased credits only15 s480p90$4.50
Seedance 2.5Purchased credits only23 s480p136$6.80
Seedance 2.5Purchased credits only10 s720p133$6.65
Seedance 2.5Purchased credits only15 s720p198$9.90
Seedance 2.5Purchased credits only23 s720p302$15.10
Seedance 2.5Purchased credits only10 s1080p233$11.65
Seedance 2.5Purchased credits only15 s1080p348$17.40
Seedance 2.5Purchased credits only23 s1080p532$26.60
MiniMax H3Purchased credits only10 s720p (768P)61$3.05
MiniMax H3Purchased credits only15 s720p (768P)89$4.45

If a video fails at any step, every credit goes back automatically. A finished video is charged again if you remake it.

See credit packs

Examples that match the template

Every example uses the same kitchen duet moves and the original take’s sound as the reference this generator follows. Results vary with your photos and the model.

Two friends in black tees

Two friends in a home kitchen; the left one strums, the right one sings the hook. 14 s with the original sound.

One golden retriever, both parts

A single dog photo used twice: one retriever strums, the other throws its head back on the “ooh”. Full 23 s with the original sound.

Plaid shirt and KC cap

A plaid-shirted strummer and a singer in a white tank and cap, the opening 9 s of the duet with the original sound.

A poodle and a grey cat

Two different pets: a poodle on the ukulele and a grey cat singing, the full 23 s with the original sound. Paws stay paws.

Where the trend comes from

Einer Bankz is known for clips of rappers singing over his ukulele. In 2018 he filmed Fredo Bang singing the hook of “Oouuh” in a small kitchen, and later posted the take on TikTok, where it has been played millions of times.

In early October 2026 it became an AI trend: people dropped themselves, friends, partners and pets into the duet, mostly with CapCut or PixVerse templates shared under tags such as #pioneertemplate and #ooouhbyfredotutorial.

Sources: the kitchen take on Einer Bankz’s TikTok · Fredo Bang – “Oouuh” (official video)

RoadsterVideo is an independent tool, not affiliated with or endorsed by Fredo Bang, Einer Bankz, CapCut, PixVerse or any tool named on this page.

What You Want Me To Do AI video generator FAQ

What are the best AI tools for making the What You Want Me To Do AI video?

Tools built for this exact trend, because general generators don’t know the clip. A template that follows the original 2018 take — like the generator on this page, OOUUH AI’s Trend mode or Starrd — keeps the kitchen, the moves and the original sound, and you only add photos. CapCut or PixVerse templates work if you can find one through TikTok tags. Kling, Runway or Veo can animate photos, but they need long prompts and still invent their own moves and music. On RoadsterVideo a video costs 15–532 credits and downloads without a watermark.

Which online generator is closest to the original trend without editing?

One that uses the original take as its motion reference and keeps its audio. Here the model follows a depth map of the first 23 seconds of the 2018 kitchen clip, and the take’s own ukulele and vocal are laid under your video, so the head-back “ooh” lands on the beat. Seedance 2.5 makes the full 23 seconds; the other models make 10 or 15. There is no timeline and no prompt.

What is the difference between a CapCut template, a dedicated generator and Kling or Runway?

A CapCut or PixVerse template is a ready-made effect someone else built: quick if you find it, but you get that template’s length and sound. A dedicated generator has the scene, the roles and the sound built in and runs in a browser. Kling, Runway and Veo are general image- and text-to-video models: flexible, but you write the scene yourself and get new moves and new music.

Is the What You Want Me To Do AI video a HeyGen-style talking avatar?

No. HeyGen, Synthesia and D-ID make one face talk or lip-sync to a script. This trend is two full-body performers side by side, one strumming a ukulele, one singing with big moves. Chatbots sometimes suggest avatar tools because they read the phrase as “make the AI video I want”.

Which tools keep the original “Oouuh” sound?

The ones that follow the original take: this page, OOUUH AI’s Trend mode and Starrd say they use the original audio. Kavel, MICT and aifredobangvideo.pro generate new audio, such as a ukulele groove or a wordless “ooh”, and general apps add none or their own. If a video has no sound, add the official “Oouuh” sound inside TikTok or Instagram.

Do I need one photo or two?

Two photos give two people: Photo 1 strums on the left, Photo 2 sings on the right. Here the same photo in both slots makes a solo duet, and pets work too. OOUUH AI’s Trend mode needs two single portraits (its Original Song mode takes one group photo); some one-photo tools replace only the ukulele player and keep the original singer. A group photo is not split, so crop each person into their own photo first.

Is “What You Want Me To Do AI” just any AI video generator?

No. Searches for the phrase return general tools such as Canva, Adobe Firefly or invideo and articles about AI that “does what you want”, but the trend is one clip: Fredo Bang singing “Oouuh” with Einer Bankz on ukulele in a kitchen in 2018. “What You Want Me To Do” is a line from that hook, not the song title.