Audio generation
Coming soon

Mirelo SFX 2.0

by Mirelo AI

Mirelo's video‑to‑sound model.
New sound for your video, same dialogue.

Mirelo SFX 2.0

Key features

Add new sound effects to my video and keep the dialogueStart creating

Specs and listening-test figures are Mirelo's own, from its October 14, 2026 launch post. The clips on this page were made on Morphic with Sonilo SFX, which does the same job today.

Hear the difference

A woman chopping carrots at a wooden kitchen counter in warm window light

Music out, dialogue kept

The original has a guitar track under her line. The rebuilt sound keeps the line and adds the knife on the board.

A man opening an umbrella beside a woman on a rainy street at night

A score swapped for the street

A piano score sits under the line. The rebuilt sound is rain on the umbrella and the street.

A man in a red jacket carrying firewood to a snowy cabin at dusk

Three shots, one place

Three generated shots stitched together. One pass gives the wind and snow one continuous sound.

A young woman diving off a wooden dock into a lake ringed by pine forest

Pop track out, splash in

A pop track covers the dive. The rebuilt sound has the run on the planks, the splash and the laugh.

A blacksmith striking a glowing bar on an anvil in a dark forge

Strikes, fire and steam

No prompt. The rebuilt sound comes from the picture alone: each hammer blow, the fire and the hiss of steam.

Technical specifications

10 min

Per run in Mirelo's studio and API, longer on request

Up to 4

Variations from one run, so you can pick the best

Kept

On by default, or off for sound effects only

81%

Over another video-to-sound model in Mirelo's blind test

Use cases

A man in a red jacket walking toward a snowy timber cabin at dusk

AI short films

Generate the shots, lock the cut, then give the whole film its sound in one run instead of fixing each clip.

A woman chopping carrots at a wooden kitchen counter in warm window light

Ads and product videos

Keep the voiceover take, drop the music the video model chose, and add your licensed track on top.

A young woman diving off a wooden dock into a lake ringed by pine forest

Social clips

Short vertical clips get the swish, the splash and the crowd they need, timed to the cut.

A woman walking across a vast vaulted stone hall in pale daylight

Long-form edits

A ten-minute edit runs in one pass, so the room tone in minute one matches the room tone in minute nine.

A blacksmith striking a glowing bar on an anvil in a dark forge

Game trailers and cinematics

Impacts, sparks and steam follow the motion, frame for frame, without a sound library or a timeline full of clips.

A cyclist riding down a wet country road through misty farmland at dawn

Video agents and platforms

Agents that stitch clips without a person checking can finish every video with clean, matching sound.

Overview

Mirelo SFX 2.0 is Mirelo AI's video-to-sound model, released on October 14, 2026. It takes a finished video and gives it new sound: effects, foley and ambience timed to what happens on screen, the spoken lines kept as performed, and any music the video model added removed. A full edit of up to 10 minutes runs in one pass.

What Mirelo SFX 2.0 does differently

Mirelo SFX 2.0 starts from the picture. Sound libraries and text-to-sound models never see your video, so every effect has to be found, trimmed and lined up by hand. Mirelo SFX 2.0 reads the action, timing and setting and builds the sound to match, around the dialogue you already have. In Mirelo's blind listening test, people preferred its sound over another video-to-sound model in 81% of comparisons (source).

Mirelo SFX 2.0 on Morphic

Mirelo SFX 2.0 is coming to Morphic soon. Until then, Sonilo SFX does the same job: open Copilot, attach your video and ask for new sound with the dialogue kept. Every clip on this page was made that way. For how to get the best result, see the Mirelo SFX 2.0 guide.

Simple pricing

Get started for free today, with the option to upgrade or cancel anytime.

Basic

$19/ month
billed as $0 per year

2400 monthly credits

1 user only

All models

Workflows

Standard

$24/ month
billed as $0 per year

3625 monthly credits

1 user only

All models

Workflows

Pro

$45/ month
billed as $0 per year

6350 shared monthly credits

1 user

+ up to 4 more at extra cost

All models

Workflows

Max

$170/ month
billed as $0 per year

24650 shared monthly credits

1 user

+ up to 9 more at extra cost

All models

Workflows

Enterprise

For higher limits

Custom

pricing and billing terms

High-volume credits
Custom seat limits
All models
Workflows
Pricing Gradient

Free

For playing around

$0

forever free

Up to 20 credits
1 user only
Limited models
Workflows
Compare plan details

FAQs

What is Mirelo SFX 2.0?
Mirelo SFX 2.0 is Mirelo AI's video-to-sound model, released on October 14, 2026. You give it a video and it returns new sound for it: effects, foley and ambience timed to the action, with the dialogue kept and any unwanted music removed. It handles a full edit of up to 10 minutes in one run.
When is Mirelo SFX 2.0 coming to Morphic?
Mirelo SFX 2.0 is coming soon. Until it arrives, Sonilo SFX does the same job on Morphic: it scores a clip up to 10 minutes long, keeps the dialogue and replaces the music and ambience. Mirelo SFX for short clips up to 10 seconds is also live.
How does Mirelo SFX 2.0 keep the dialogue?
It finds the speech in your clip, keeps it, and builds new effects and ambience around it, lowering them under the voices. Speech is kept by default. Turn it off when you want sound effects only, for example before you record your own voiceover.
Can Mirelo SFX 2.0 remove music from a video?
Yes. Everything in the original soundtrack except the speech is replaced, so music a video model added without being asked is gone from the result. You can then add your own music on top.
How long a video can Mirelo SFX 2.0 handle?
Up to 10 minutes per run in Mirelo's studio and API, according to Mirelo. Longer videos are available through Mirelo's Enterprise tier. Because the whole edit runs at once, the sound stays consistent from the first shot to the last.
Do I need to write a prompt?
No. The video is the main input, and most clips need nothing else. A short prompt helps when you want a specific sound, such as heavy rain or a crowd far away, and a short list of sounds to leave out keeps things like music or wind out of the result.
What are the three versions of Mirelo SFX 2.0?
Video-to-sound is the main model, and the one people mean by Mirelo SFX 2.0. Text-to-sound makes a sound from a prompt when it does not need to match the picture, such as a background ambience. A stem version that makes one synced sound at a time is announced as coming soon.
Is Mirelo SFX 2.0 better than a video model's own audio?
In Mirelo's own blind listening test, people preferred Mirelo SFX 2.0's sound over the native audio of every video model it tried, on cinematic clips. Results will vary by clip, so listen to both before you replace a soundtrack you already like.
Were the clips on this page made with Mirelo SFX 2.0?
No. Mirelo SFX 2.0 is not on Morphic yet, so every clip was made on Morphic with other models: the video with Morphic H3 and Seedance 2.0, the new sound with Sonilo SFX. On the music examples, a music bed made on Morphic was added under the dialogue first, to show it being removed.

More about Mirelo SFX 2.0

Guides, prompts, and comparisons for Mirelo SFX 2.0.

Mirelo SFX 2.0 guide

How to use Mirelo SFX 2.0: keep the dialogue, remove unwanted music, sync effects to the action and keep one sound across a full edit.

Other models

Explore the rest of the Morphic model catalog.

See all models →

OmniHuman 2.0

ByteDance

ByteDance's audio-driven human video model. One photo, one track, five minutes in one take.

Fish Audio Drama 3

Fish Audio

Fish Audio's most expressive voice model. Direct the performance, down to the word.

Vidu Q4

Shengshu Technology

Shengshu's latest video model. Up to 12 reference images and 3 voices held consistent across one clip, with native audio from 540p to 4K.

Nano Banana 2.1

Google DeepMind

Google DeepMind's new Flash image model. Better design, cleaner edits, steadier characters.