Narrate a Claude Motion explainer with Sume TTS and Timeline: 12 cents

You can narrate a silent Claude Motion MP4 for about $0.12: generate the voiceover with Sume text to speech, then render the video and the audio together with Timeline. A 450-character script is 450 / 1,000 x $0.0475 = $0.021 of speech, and a Timeline render of up to one minute is $0.10, so the total is $0.121.

One caution first. Anthropic's announcement of Claude Motion does not say whether the exported MP4 includes an audio track. This post assumes you have a picture-only file, or that you want to replace the audio.

The steps

• Write the script from the animation script Claude already produced, so the words match the numbers on screen.
• Generate speech. Sume lists TTS at $0.0475 per 1,000 characters, with a maximum request of 20,000 characters.
• Import the Motion MP4 into Sume storage with POST /v1/media-imports, because Timeline sources must be on media.sume.com.
• Call POST /v1/timeline-1.0/render with the video slot and the audio. Include an Idempotency-Key, which the docs require.
• Poll the job and read video_url from the result.

Cost table

Check the length before you render

Timeline bills by output minute, rounded up, so a 61-second file costs two minutes. Spoken English at a moderate pace is roughly 150 words a minute, and 450 characters is about 75 words, so about half a minute of speech. If the audio runs longer than the picture, trim the script rather than paying for a second minute. The Timeline docs also list a free POST /v1/timeline-1.0/plan call that compiles the request and returns the duration without billing.

The audio field takes duration_seconds between 1 and 1,800, and a render can hold up to 200 video slots. A 30 to 90 second explainer is far inside both limits.

A quarter of weekly explainers

For 13 weekly videos with one render each, the cost is 13 x $0.121375 = $1.577875, which rounds to $1.58. If every episode needs one extra render because the animation changed after narration, each episode is $0.021375 + 2 x $0.10 = $0.221375, and 13 of them cost $2.877875, about $2.88.

Either figure is small next to the time spent on the script. The cost that deserves attention is the number of renders, since the Timeline fee is charged per render attempt and TTS only per character.

Where this does not fit

If the animation changes after you narrate it, the voice no longer matches. Anthropic says you can change any word, number or timing, so lock the animation before you generate speech. Re-running TTS for a changed script is cheap, at $0.0475 per 1,000 characters, but the Timeline render bills again for each attempt.