A 30-minute voiceover: the 1,800 s Timeline cap and two TTS jobs
A 1,800-second Timeline render costs $3.00 and a 30-minute read exceeds the 20,000-character TTS cap, so it takes two jobs. Per-vendor math, read 2026-10-08.

The longest voiceover a single Timeline render can carry is 1,800 seconds (30 minutes), which renders for 30 x $0.10 = $3.00, and a 30-minute script is too long for one Sume TTS request, which is capped at 20,000 characters. At an assumed 900 characters of script per minute, 30 minutes is 27,000 characters, so it takes two TTS jobs and $1.28 of TTS at the Sume rate.
The characters at three reading rates
The characters-per-minute values are assumptions you should replace with your own: 750, 900 and 1,050. The TTS columns are list price per character with no job floor, because every job here is far above 211 characters. Jobs = ceil(characters / 20,000) for Sume, whose catalog states that maximum; I did not read a request cap on Microsoft's page and give none.
| Chars per minute (assumed) | Characters | Sume TTS jobs | Sume TTS | MAI-Voice-2.1 | MAI-Voice-2.1-Flash |
|---|---|---|---|---|---|
| 750 | 22,500 | 2 | $1.0688 | $0.495 | $0.3375 |
| 900 | 27,000 | 2 | $1.2825 | $0.594 | $0.405 |
| 1,050 | 31,500 | 2 | $1.4963 | $0.693 | $0.4725 |
The cap and what to do at the edge
Timeline accepts audio.duration_seconds from 1 to 1,800, and render billing is $0.10 per started minute with a 30-minute maximum (Timeline 1.0). A 31-minute piece cannot be rendered in one job. The usual fix is two renders and a join elsewhere, or a cut that fits. Within the cap, splitting the script at a paragraph near the middle gives two TTS requests that each fit under 20,000 characters, and you can join the two audio files with Timeline audio at $0.01 before the render.
Gemini's output charge is by time, not characters: 30 minutes is 30 x 60 x 25 = 45,000 audio tokens, and 45,000 x $9.00 / 1,000,000 = $0.405 at the current Flash output price, before text input (Gemini pricing). That figure does not depend on how many characters you read.
Totals for the 900-character case
Sume: TTS $1.2825 + one $0.01 join + $3.00 render = $4.2925. With MAI-Voice-2.1 audio imported instead of Sume TTS: $0.594 + $0.01 + $3.00 = $3.604, and the render dominates either way. For a long voiceover the Timeline render, not the speech, is the largest line.
Shorter cuts of the same script
If the render is the biggest line, cut the length. Every minute removed from the output saves $0.10, so trimming a 30-minute piece to 24 minutes saves $0.60 on the render ($2.40 instead of $3.00) and, at 900 characters a minute, 5,400 characters of speech, which is $0.2565 on Sume. Both savings come from the same editorial decision, so make it before synthesis, not after.
The three assumed rates matter more than the vendor choice. Between 750 and 1,050 characters a minute the Sume TTS bill moves from $1.07 to $1.50, a spread of $0.43, while the render is $3.00 in both cases.
One more number for planning: 30 minutes at 900 characters a minute is 27,000 characters, so the script splits into two requests (for example 13,500 and 13,500), each well under the 20,000-character cap, and the two outputs join with one $0.01 concat before the render.
Sources
Related posts
More in Use cases
- A 52-minute podcast on Sume: the 1800-second limit and the split
Sume trim and inspect refuse sources over 1800 seconds. A 52-minute episode must be split into two 26-minute parts first; six clips then cost about $2.50.
- 90-second explainer from fifteen 6-second scenes, plus the render
Fifteen 6 s scenes at 720p cost $11.25 on Wan 3.0 or Omni and $52.00 on Seedance 2.5; the 90 s timeline render adds $0.20 (2 billed minutes at $0.10).
- Ad hook test matrix: 4 hooks x 2 ratios on Omni Flash, priced
Four ad hooks in 16:9 and 9:16 as 5-second Omni Flash drafts cost $1.50 at 360p on Sume; re-rendering two winners at 1080p adds $1.875. The full matrix.
- Adobe's Cyber 5 forecast: plan five daily Sume bulk queues
Adobe projects $47.5B for Cyber 5, 17.3% of holiday online sales. Map the five days to five Sume bulk queues with one idempotency key each and a cap per item.
Written by Sume