Quick audio & asset tools

Audiograms, stem splitting, lyric videos, logo animations, and GIF export — small catalog tools for fast turnarounds.

Last updated August 23, 2026

Not everything needs a full multi-scene production. These catalog workflows and file tools turn one input into one useful asset. Find the workflows in Discover or ask for them by name in chat.

Audiogram

audiogram turns an audio file into a shareable waveform video — the animated-waveform card you see for podcast clips and voice notes. Attach the audio (or point at one already in Files), run, and download the rendered video. The waveform is composited in Remotion, so it is a real video file ready for any platform.

Stem split

stem-split separates a mixed track into vocals and instrumental stems using a source-separation model (MelBandRoFormer). Attach a song, get both stems back as audio files in your library. Useful for building a clean instrumental bed, isolating a vocal, or feeding the instrumental into a lyric video.

Lyric video

lyric-video produces a karaoke-style video for an original song: generated visuals, an instrumental music bed, and word-synced captions driven by TTS word timings — each word lights up as it lands. Describe the song and mood in the brief; captions are always on for this workflow.

Note this builds around an original performance — see how named-song requests are handled (style facsimile, never a copy of the recording).

Logo animation

logo-animation renders a short logo-reveal video (a Remotion composition) from your real logo. It requires an actual logo image — from the workflow input, an attached file, or your brand kit, in that order. If none is found the run fails up front rather than inventing a lookalike logo. The render is billed as a compose render, like the assemble step of a video.

GIF export

Any settled video in your library can be exported as an animated GIF:

POST /files/{fileId}/gif   { startS?, durationS? }

It is a free transform (no generation credits — the GIF only counts against storage), defaults to the first 30 seconds, and lands as a sibling file in the same folder. In the UI it is the Export GIF action on a video tile in Files and on the outputs card. See Exporting & sharing.