Re-record one voiceover line: TTS cost for 50 to 5,000 characters

Fixing one 120-character line costs $0.0057 in Sume TTS, 12% of redoing a 1,000-character script at $0.0475. Gluing it in is free inside a Timeline render.

5 min readSume
All posts

Re-recording a single 120-character line with Sume text to speech costs $0.0057. Re-recording the whole 1,000-character script costs $0.0475. Because TTS is billed by character, the cost of a fix scales with the size of the fix, so re-record the line, not the script, and join it back in.

The join is the part that can cost more than the line. Sume's standalone timeline audio job is a flat $0.01, which is more than the line itself. Inside a Timeline render the join comes free.

What does each size of re-record cost?

The Sume TTS rate is $0.0475 per 1,000 characters (from the pricing package, matching the catalog on the Sume API pricing page). The catalog notes that spaces and punctuation count toward the character total. The table prices a range of fix sizes and compares each to a full 1,000-character script.

TTS cost by characters re-recorded at $0.0475 per 1,000 characters, computed (read 2026-10-03)
CharactersCostShare of a 1,000-character script
50$0.0023755.0%
120$0.005712.0%
300$0.0142530.0%
500$0.0237550.0%
1000$0.0475100.0%
5000$0.2375500%

What does it cost to put the fixed line back?

Per the Timeline audio docs, a concat job joins up to 20 hosted audio parts into one gapless file with no re-TTS. That job is a flat $0.01, per the pricing package. If the audio is going into a render anyway, the Timeline 1.0 docs say a join that is only needed inside one render belongs in audio.parts[] and the separate job can be skipped.

Three ways to fix one 120-character line in a 1,000-character script, computed (read 2026-10-03)
ApproachTTS costJoin costTotal
Re-record the whole script$0.0475none$0.0475
Re-record the line, join with a timeline audio job$0.0057$0.01$0.0157
Re-record the line, join inside the render with audio.parts[]$0.0057none$0.0057

When is re-recording the whole script the better choice?

Splicing assumes the new line sits naturally between its neighbors. Three situations favor redoing the script despite the higher cost:

  • The change is a voice or style change that applies to every line.
  • The edit changes the rhythm of the sentences around it, so a seam would be audible.
  • The saving is smaller than the time you spend splicing; at under five cents a script, a human hour is the real cost.

How do you keep fixes cheap from the start?

Generate the voiceover in segments from the beginning, one line or one paragraph per TTS call, and keep the returned audio URLs. A later fix then replaces one entry in the audio.parts[] list instead of forcing a regeneration. Keep the list order in your own record and reuse the same idempotency key only for a retry of the same line, not for a new take of it.

Putting the cents in context

At $0.0475 per 1,000 characters, re-recording a 120-character line costs about $0.0057, under a cent. Re-recording a whole 2,000-character script costs $0.095. The money is not the reason to patch one line. The reasons are consistency of voice across the take and the time it takes to splice a replacement into the finished audio, so make that choice on workflow, not on price.

If you plan to patch lines routinely, keep the script split into the same line units you will re-record, with one request per line, so a fix never touches neighbouring text.

Sources

Related posts

More in Pricing

All Pricing posts

Written by Sume