aivideo.zone Runway MCP adds sound effects, voiceovers and background removal to AI video agents
Runway changelogRunway's product changelog, with the MCP updates of October 2026
Runway MCP is the server that lets an AI agent, in Cursor or any other client that speaks the Model Context Protocol, call Runway's generation tools directly. In the week of 12 October it learned to make sound.
Mirelo SFX 2.0, 15 October
Give the agent a clip and describe the sound you want, from footsteps to room ambience, and Mirelo SFX 2.0 returns it as a separate audio track, ready to drop into the edit rather than baked into the video.
Fish Audio Drama 3, 14 October
Fish Audio Drama 3 turns a script into a voiceover. It has seven voices and takes delivery cues for whispers, laughter and dramatic moments.
BiRefNet v2, 15 October
Background removal with BiRefNet v2: it finds the foreground on its own, for product shots, illustrations and assets headed for a composite.
Also this month
Seedream 5.0 Flash (12 October) makes 1K or 2K images from text or from up to ten reference images, also in Runway MCP. In Runway's own Agent, Seedance 2.5 Draft mode (2 October) makes fast 480p drafts. Enterprise workspaces can now cap the credits each project spends (9 October).
Who gets it
The three MCP additions are on paid plans.
The separate audio track is the detail that matters. A video model that generates sound with the picture gives you a mix that is hard to fix; an effects track on its own can be levelled, cut and replaced in the edit like any recorded foley. Put that together with a scripted voiceover from the same agent and the rough cut of a short spot, picture, effects and voice, can come out of one conversation, with every layer still editable. Check the credit cost per generation on your plan before letting an agent loop on sound, since an agent will happily ask for take after take.
Runway MCP adds sound effects, voiceovers and background removal to AI video agents: questions
What is Runway MCP?
A Model Context Protocol server that lets an AI agent use Runway's tools from an MCP client such as Cursor.
What does Mirelo SFX 2.0 do?
It generates sound effects for a clip from a text description and returns them as a separate audio track.
How many voices does Fish Audio Drama 3 have?
Seven, with delivery cues for whispers, laughter and dramatic moments.
Are the new tools free?
No. Mirelo SFX 2.0, Fish Audio Drama 3 and BiRefNet v2 are listed for paid plans.
Source: Runway changelog checked against the source
More AI video news
LTX-2 v1.5.0
Lightricks released v1.5.0 of the LTX-2 code that runs the open LTX-2.5 AI video model on 18 October 2026: image conditioning now says whether a frame is replaced or only guides, keyframe interpolation moved into the main two-stage pipeline, and fp8 matrix maths speeds up GPUs of compute capability 8.9 and newer. Outputs are no longer bit-identical to earlier releases.
Tavus Griffin-Lite
Tavus has shown Griffin-Lite, a research preview of what it calls a Human Interaction Model: full-duplex video-to-video, an AI video face that listens and answers at 720p in real time, averaging 0.43 seconds of response on an H100. In its study 48 per cent of participants took it for a human. Waitlist only.
Helix World preview
Noiz AI and HKUST have released Helix World preview, an interactive audio-visual world model: 768×512 AI video at 24 frames a second generated live on a single H800 as you steer with WASD or the arrows, with 48 kHz spatial stereo sound. Built on LTX-2.3, weights on Hugging Face, demo at helixworld.org.