Mirelo AI Sfx 2.0 Video To Video API Documentation

Mirelo AI Sfx 2.0 Video To Video API Documentation

Playground

Mirelo SFX 2.0 Video-to-Video adds synchronized AI-generated sound effects to an input video and returns the original video with its new soundtrack, preserving the source visuals while enhancing sound design for film, social content, games, ads, and post-production workflows. Ready-to-use REST inference API, best performance, no coldstarts, affordable pricing.

Features

Mirelo SFX 2.0 Video-to-Video adds synchronized sound effects and ambience to an existing video while keeping the original visual footage. The model analyzes the action in the source video, generates matching audio, and returns the video with its new soundtrack.

Use optional text guidance to describe the sounds you want, add a negative prompt to exclude unwanted audio elements, and preserve existing speech when dialogue should remain in the final soundtrack.

Why Choose This?

• Video-synchronized sound generation
  Generate sound effects and ambience that follow the timing and action of the source footage.

• Original visuals preserved
  Add generated audio without creating or modifying the video frames.

• Prompt-guided sound design
  Describe the desired sound effects, environment, atmosphere, and audio direction.

• Negative prompt control
  Specify sounds, voices, music, or other elements that should be avoided.

• Speech preservation
  Keep existing dialogue while generating a new surrounding soundscape.

• Up to 10 minutes
  Request synchronized sound generation from 1 to 600 seconds.

Parameters

How to Use

1. Provide a source video — Upload a video or supply its URL.
2. Add sound guidance optional — Describe the effects, ambience, and soundtrack direction in prompt.
3. Add exclusions optional — Use negative_prompt to specify sounds that should be avoided.
4. Configure speech preservation — Keep preserve_speech=true when existing dialogue should remain.
5. Set duration — Choose the requested sound-generation length from 1 to 600 seconds.
6. Set seed optional — Use a fixed seed when reproducibility is needed.
7. Submit — Generate and retrieve the video with its synchronized soundtrack.

Pricing

Pricing is based on the requested duration.

The rate is $0.016 per requested second.

Billing uses the requested duration.

If the source video is shorter than the requested duration, sound generation stops at the end of the source video, but billing still uses the requested duration. Set duration no longer than the usable source clip when additional requested time is unnecessary.

Best Use Cases

• Film and cinematic sound design — Add synchronized ambience, impacts, movement, and environmental effects to footage.
• Gameplay videos — Generate action-matched effects for movement, combat, objects, and environments.
• Product videos — Add mechanical, interaction, and environmental sounds to product demonstrations.
• Social media content — Create richer sound design for short-form visual content.
• Foley-style workflows — Generate footsteps, impacts, object handling, fabric movement, and other action-based effects.
• Alternate soundtrack exploration — Test different sound directions without changing the source visuals.
• Speech-preserving sound design — Keep dialogue while rebuilding the surrounding soundscape.

Pro Tips

• Describe concrete sound sources, timing, intensity, movement, and environment.
• Explain how sounds should correspond to visible actions when synchronization matters.
• Use negative_prompt to reduce unwanted music, voices, or background elements.
• Keep preserve_speech=true when source dialogue should remain in the final video.
• Set preserve_speech=false when you want the generated soundtrack focused on sound effects only.
• Match duration to the usable source-video length to avoid paying for requested time beyond the available footage.
• Start with shorter generations when testing a new sound direction.
• Keep prompts focused when precise synchronization is more important than a complex soundscape.

Notes

• video is required.
• The source video must be at least 1 second long.
• duration supports 1–600 seconds and defaults to 10.
• Sound generation cannot continue beyond the source video length.
• Billing still uses the requested duration when it exceeds the available source-video duration.
• preserve_speech defaults to true.
• preserve_speech=true preserves existing speech, not all original background audio.
• Generation longer than 10 seconds is processed in consecutive windows.
• The endpoint returns the source video with generated audio rather than standalone audio.

Related Models

• Mirelo SFX 2.0 Text-to-Audio — Generate sound effects and ambience directly from text descriptions.
• Mirelo SFX 2.0 Video-to-Audio — Generate synchronized standalone audio from an input video.
• Mirelo SFX 2.0 Video-to-Video — Add synchronized generated sound effects to an existing video and return the completed video.

Authentication

For authentication details, please refer to the Authentication Guide.

API Endpoints

Submit Task & Query Result

set -euo pipefail

export WAVESPEED_API_KEY="your-api-key"

REQUEST_BODY=$(cat <<'JSON'
{
  "video": "https://interactive-examples.mdn.mozilla.net/media/cc0-videos/flower.mp4",
  "preserve_speech": true,
  "duration": 10
}
JSON
)

# 1. Submit the prediction.
SUBMIT_RESPONSE=$(curl --silent --show-error --fail-with-body \
  -X POST "https://api.wavespeed.ai/api/v3/mirelo-ai/sfx-2.0/video-to-video" \
  -H "Authorization: Bearer ${WAVESPEED_API_KEY}" \
  -H "Content-Type: application/json" \
  -d "${REQUEST_BODY}")

TASK=$(printf '%s' "${SUBMIT_RESPONSE}" | jq 'if type == "object" and has("data") then .data else . end')
PREDICTION_ID=$(printf '%s' "${TASK}" | jq -r '.id // empty')
if [ -z "${PREDICTION_ID}" ]; then
  printf 'Submission response did not contain a prediction id
' >&2
  exit 1
fi
RESULT_URL="https://api.wavespeed.ai/api/v3/predictions/${PREDICTION_ID}/result"

# 2. Poll until the prediction finishes.
while true; do
  RESPONSE=$(curl --silent --show-error --fail-with-body \
    "${RESULT_URL}" \
    -H "Authorization: Bearer ${WAVESPEED_API_KEY}")
  RESULT=$(printf '%s' "${RESPONSE}" | jq 'if type == "object" and has("data") then .data else . end')
  STATUS=$(printf '%s' "${RESULT}" | jq -r '.status // empty')

  case "${STATUS}" in
    completed) printf '%s\n' "${RESULT}" | jq '.outputs'; break ;;
    failed|cancelled|timeout|deleted) printf '%s\n' "${RESULT}" | jq . >&2; exit 1 ;;
    *) sleep 2 ;;
  esac
done

Parameters

Task Submission Parameters

Request Parameters

Response Parameters

Result Request Parameters

Result Response Parameters