Sume TTS cost by length: 1 to 20,000 characters, rounded up

Sume TTS 1.0 costs $0.0475 per 1,000 characters of transcript, and each job is rounded up to the next whole cent. A 500-character line bills 3 cents, a 1,000-character script bills 5 cents, and the 20,000-character maximum bills 95 cents.

The per-character rate is the provider list price of 38 micro-dollars per character times Sume's 1.25 margin. The rounding is why many short jobs cost more than one long one.

The ladder

Raw cost is characters times $0.0000475. The billed amount is that rounded up to the cent.

Where the rounding bites

Up to 210 characters, every job bills the 1-cent minimum, which is $0.05 per 1,000 characters at that size, a little above the headline rate. Cutting a 2,000-character script into twenty 100-character lines turns a 10-cent job into twenty 1-cent jobs, or 20 cents. That is the cost of getting twenty separate files.

If you only need separate lines for editing, use one job with sentence slices instead. The reference describes segmentation.mode=sentence, which needs word timestamps and, for audio per slice, a wav or raw output.

Limits that shape the price

The same reference sets the ceilings your estimate has to respect.

• A request takes at most 20,000 characters.
• Audio longer than 1,200 seconds fails with tts_duration_exceeded.
• Billing is by transcript characters, so speed settings change the length of the audio, not the price.
• Cost is reserved when the job is queued, and a request with an invalid voice id is refused with 400 invalid_voice_id before credits are reserved.

A quick way to price a script

Count the characters including spaces, multiply by 0.0000475, and round up to the next cent. For a batch of different lengths, round each job before you add them. The jobs and results guide explains how to read the final cost from a finished job.