mcp.film / Sound Design & SFX / Mirelo MCP
Mirelo MCP
Foley generated against the picture, then extended or patched region by region.
What it does
Mirelo's official remote MCP exposes its SFX V1 video-to-audio model: hand it a clip and it reads the on-screen action, motion and pacing and generates effects synced to what it sees, with an optional text prompt to steer the character of the sound. Text-to-sound covers the cues with no plate, and two repair tools matter more than they sound — extend a bed that came back short, or inpaint one region of a clip while leaving the rest untouched, so a single bad hit does not cost the whole pass. A preflight estimate returns credits and runtime before it generates, which is what makes batching a folder of shots safe.
Connect
https://mcp.mirelo.ai/mcp
claude mcp add --transport http mirelo https://mcp.mirelo.ai/mcp
{
"mcpServers": {
"mirelo": {
"command": "npx",
"args": [
"-y",
"mcp-remote",
"https://mcp.mirelo.ai/mcp"
]
}
}
}
{
"mcpServers": {
"mirelo": {
"url": "https://mcp.mirelo.ai/mcp"
}
}
}
Field notes
Announced 30 Jul 2026; hosted endpoint still answering with a 401 challenge on 3 Oct 2026, install command unchanged. Billed against the same Studio credits as the Mirelo web app, so it needs an account with balance — the preflight tool is worth calling first. Limits Mirelo states outright: music generation and multi-stem generation are not exposed in MCP v1, and result download links are temporary, so an agent that generates without saving loses the take. In Claude (web and Desktop), file uploads fail until you allowlist studio-project-assets.s3.eu-central-1.amazonaws.com under Settings → Capabilities — the usual reason a video-to-SFX call never sees the clip. Documented clients are Claude, Claude Code, Cursor, Codex and ChatGPT; Codex has its own `codex mcp add mirelo --url https://mcp.mirelo.ai/mcp` plus `codex mcp login mirelo`. Tool names are not published — call tools/list. Closest comparisons here are ElevenLabs text_to_sound_effects and Sonilo's video_to_sfx; Mirelo is the narrower specialist, and region-level inpainting is the thing neither of those offers.
Also in Sound Design & SFX
Local Stable Audio 3 for temp score and SFX: generate, inpaint, extend, loop and loudness-normalize offline.
Local, offline sound-effect generation with Meta's AudioGen — no API key, no credits.
Loudness, true-peak and voice metering that grades a mix against real delivery specs.
Drives a live REAPER session: tracks, MIDI, FX, mixing, stems, and loudness analysis.
Search freesound.org's CC-licensed sound library by natural language for SFX pulls.
Auphonic's automated audio post (loudness, leveling, denoise) reachable through Zapier MCP.
Pairs well with
Search a catalogue of real recordings and buy a lifetime worldwide sync licence per track, WAV plus certificate.
Hosted MCP for TTS, voice cloning, music, stem and speaker separation, and denoise.
Official Cartesia (Sonic) server: low-latency TTS, cloning, voice change, localization.