Vidu Q3 और Q3 Pro मॉडल पर 50% छूट · केवल WaveSpeedAI | 20 मई – 2 जून

Mirelo SFX1.6 Inpaint Audio API

mirelo-ai /

Mirelo SFX1.6 Inpaint Audio is a fast AI audio inpainting model that regenerates a selected segment of a short audio clip while preserving the audio outside that region. Ready-to-use REST inference API for sound effect repair, audio cleanup, segment regeneration, game audio, video production, cinematic sound design, and professional audio editing workflows with simple integration, no coldstarts, and affordable pricing.

audio-to-audio
Input

Drag & drop करें या upload के लिए click करें

Idle

$0.01per run·~100 / $1

ExamplesView all

Related Models

README

Mirelo AI SFX 1.6 Inpaint Audio

Mirelo AI SFX 1.6 Inpaint Audio regenerates a selected segment inside an existing audio clip while preserving the audio before and after that region. It is designed for audio repair, sound replacement, creative audio editing, and localized sound redesign workflows where you want to rewrite only part of a clip instead of regenerating the whole file.

Why Choose This?

  • Segment-level audio editing Regenerate only a selected section of an audio clip while keeping the rest unchanged.

  • Prompt-guided inpainting Add an optional text prompt to guide how the selected region should be regenerated.

  • Multiple variations Generate up to 4 inpainted variations in one request with num_samples.

  • Preserves surrounding audio Audio outside the selected inpaint segment remains intact.

  • Production-ready workflow Useful for sound repair, effect replacement, music cleanup, and creative segment editing.

Parameters

ParameterRequiredDescription
audioYesSource audio clip. Maximum duration is 10 seconds. Audio outside the selected segment is preserved.
start_sYesStart of the inpaint region in seconds. Must leave a 1 second left gap. Default: 2.
end_sYesEnd of the inpaint region in seconds. The segment span must be 1–8 seconds. Default: 6.
promptNoOptional text prompt guiding the regenerated segment.
num_samplesNoNumber of variations to generate. Range: 1–4. Default: 1.

How to Use

  1. Upload your audio — provide the source audio clip you want to edit.
  2. Set the inpaint region — choose start_s and end_s to define the segment you want to regenerate.
  3. Add a prompt (optional) — describe the sound, mood, or style you want for the regenerated segment.
  4. Set number of samples — choose how many variations you want, from 1 to 4.
  5. Submit — run the model and download the inpainted audio results.

Example Prompt

Replace this section with a darker cinematic ambience, soft low rumble, subtle tension, and cleaner transition into the following audio.

Pricing

Pricing is based on the inpainted segment length and number of samples.

Inpaint Length1 Sample2 Samples3 Samples4 Samples
1s$0.01$0.02$0.03$0.04
2s$0.02$0.04$0.06$0.08
4s$0.04$0.08$0.12$0.16
6s$0.06$0.12$0.18$0.24
8s$0.08$0.16$0.24$0.32

Billing Rules

  • Pricing is $0.01 per generated second
  • prompt does not affect pricing
  • Pricing is based on the regenerated segment only, not the full uploaded audio duration

Best Use Cases

  • Audio repair — Replace damaged or unwanted sections inside a clip.
  • Sound redesign — Change one localized segment without touching the rest of the recording.
  • Music cleanup — Rewrite small regions in intros, drops, transitions, or effects.
  • Creative editing — Test alternate versions of a section with multiple generated samples.
  • Loop and transition refinement — Improve a specific transition area inside a short clip.

Pro Tips

  • Choose the smallest region that needs replacement for more controllable results.
  • Use a clear prompt only when you need stronger direction for the regenerated section.
  • Keep the source audio clean so the transition into and out of the edited region feels smoother.
  • Increase num_samples when you want to compare multiple replacement options.
  • Double-check the start and end boundaries so the model edits exactly the intended segment.

Notes

  • audio, start_s, and end_s are required.
  • Source audio duration must not exceed 10 seconds.
  • The inpaint segment span must be between 1 and 8 seconds.
  • start_s must leave at least 1 second of audio before the edited region.
  • Audio outside the selected inpaint region is preserved.
  • Pricing depends on the regenerated segment length and sample count.

Related Models

  • Mirelo AI SFX 1.6 Extend Audio — Extend an existing audio clip with newly generated continuation.
  • Other Mirelo AI sound generation workflows — Useful when you need full audio generation instead of localized editing.
Accessibility:This website uses AI models provided by third parties.