This site’s voiceover and music generation are not currently available. You can still plan audio and work with tracks you already own or have permission to use in a local project. This guide primarily describes the open-source Claude Motion toolkit and Remotion; an optional external voice provider is a separate service, not an included site capability.
Write speech to fit the scene
Start with a short script that adds meaning instead of reading every visible word. For a proposed 15-second example, reserve 0–1 seconds for the opening, 1–11 seconds for two spoken sentences, and 11–15 seconds for the final action and music tail. Read or record the script and measure its actual length. If it overruns, remove words or adjust the storyboard. Do not force it into the slot by making it unnaturally fast. Keep the final sentence’s ending clear of the scene cut.
Put all sound on one timing plan
Maintain a cue list with track filename, start time, end time, role, and intended level. For example, narration begins at 1 second, a card click occurs at 4 seconds, and music fades during the final second. These times are editorial choices, not automatic synchronization guarantees. In Remotion, keep audio files under public and reference the correct local paths. The toolkit also includes procedural sound tools and a shared timing workflow. Do not confuse those generated effects with a recorded human narration or assume an optional provider has already been configured.
Make speech the reference for the mix
First listen to narration alone and remove distracting gaps or abrupt cuts. Then introduce music quietly and reduce it further during important words. Remotion supports fixed volume and changes over time; a value such as 0.15 can be a starting experiment for a music track, not a universal loudness target. Different recordings have different levels. Listen for clipping, music masking consonants, and a final note cut in half. Fade at sensible boundaries, and compare the full mix on headphones and a phone speaker before exporting.
Keep permissions with the audio files
Record the source, creator, license terms, required credit, and intended platforms for every track. YouTube’s Audio Library distinguishes tracks that require attribution from those that do not; copy the specified credit when required. Availability in one platform’s library does not automatically establish permission for every other platform or advertisement. For a generated voice, use a voice you are authorized to use and check the provider’s terms. If rights are uncertain, replace the track or deliver a silent version while you resolve them.
Verify the exported file, not only the preview
Render a new file after an audio change and open the MP4 in an ordinary player. Check that the intended track exists, speech starts on time, music stops cleanly, and muting still leaves the message understandable. On this site, shared visual HTML and browser previews are silent; an audible deliverable requires the local video workflow. Local track paths accepted by its CLI use assets/ names such as assets/narration.wav, not the public/ convention of Remotion. Keep those two project formats separate and follow the instructions in the project you actually downloaded.
Your next step
Treat audio as timed source material with its own permissions and review pass. Clear speech, controlled music, and a verified final file matter more than adding sound to every movement.
Explore the workspaceSources & further reading
References link to original documentation and creators. Product availability can change; check the linked source before subscribing or installing.
- Claude Motion toolkitwhaleyxbt
- Importing assetsRemotion
- Controlling volumeRemotion
- YouTube Audio LibraryYouTube
- Claude Motion Skills public CLIClaude Motion Skills
- Claude Motion Skills service availabilityClaude Motion Skills