$40 a week for text to speech: how many characters on Sume

A $40 weekly text-to-speech budget on Sume covers 842,105 characters at $0.0475 per 1,000 characters ($40 / $0.0475 x 1,000). In practice that is 255 narrations of 3,300 characters, which cost $0.15675 each and total $39.97, or 42 maximum-length requests of 20,000 characters at $0.95 each.

The formula

Sume TTS 1.0 is priced per transcript character: $0.0475 per 1,000 characters, which is the provider list price times 1.25. Spaces and punctuation count. The maximum per request is 20,000 characters, which is $0.95.

Cost per request = characters / 1,000 x $0.0475. For 3,300 characters that is 3.3 x $0.0475 = $0.15675. Divide your budget by that: $40 / $0.15675 = 255.2, so 255 requests.

Counting characters before you send

Do not estimate by word count alone. The price follows characters including spaces and punctuation, so a script with long words or lots of quotation marks costs more than the same word count with short words. Count the string you actually send.

A useful habit is to compute the cost in code before submitting. len(text) / 1000 x 0.0475 gives the price in dollars. If it exceeds your per-narration limit, trim or split the script.

• Maximum 20,000 characters per request
• Minimum charge of 1 cent per request in the catalog
• Spaces and punctuation count

What the weekly budget looks like in scripts

At 3,300 characters per narration, 255 narrations a week is about 36 a day. If you only need 10 a day, the same budget leaves $29.03 of headroom per week for retakes: 70 narrations cost $10.97.

If your scripts average 9,000 characters (long product explainers), the budget drops to 93 narrations at $0.4275 each, or $39.76 in total. Splitting a 20,000-character script into two requests costs the same as one request, since the price is linear.

A second scenario

Turn the weekly budget into a monthly one. $40 a week is about $171.43 over 30 days ($40 / 7 x 30), which covers roughly 3.6 million characters (842,105 x 30 / 7 = 3,609,021). If your real volume is far below that, you can lower the cap; if your scripts grow, the cost scales in a straight line with characters, with no step changes.

Edge cases

The catalog also publishes an estimate of 5 cents for a 1,000-character request. That is a rounded hold for the balance check, not a different price. If a request fails, the catalog's reservation policy refunds it before capture.

For other audio costs in the same pipeline, such as music at a flat $0.125 per generation, see the related posts below.

Reproduce the numbers

Open https://api.sume.com/v1/catalog, which needs no key, and find the entry for the product. Each model_pricing row carries price, unit, pricing_basis and the estimated, minimum and maximum cents. Multiply the unit price by your quantity and round only at the end. The catalog also has applied_* fields that include an agent fee of 550 basis points; the figures in this post use the base prices.