What does a voiceover cost on Sume? $0.0475 per 1,000 characters

Sume text to speech lists at $0.0475 per 1,000 characters, so a 600-character script costs about 2.9 cents and a 10,000-character script costs $0.475. The price is linear in characters, which makes a voiceover one of the cheapest lines in a video budget.

This page turns the rate into a table you can budget from, explains what the reserve shows in the catalog and gives a small function for planning. The rate is the catalog price read on 2026-10-02, and it applies to the router and the 1.0 TTS entry alike.

Because the price is per character, there is no minimum length worth worrying about on the billing side. A one-line hook and a paragraph differ only by the arithmetic.

How much does a script of a given length cost?

Divide the character count by 1,000 and multiply by $0.0475. The table assumes the count is the length of the text you send, with no extra markup added by you.

Word counts in the last column are rough guides at about five characters a word including spaces. Count your own text, because punctuation and long words move the number.

What reserve will the wallet hold?

Before the audio is made, Sume holds an estimate. For TTS the catalog shows an estimate of 5 cents, a minimum of 1 cent and a maximum of 95 cents, so a very long request needs the wallet to cover up to 95 cents at submit time.

The reserve is not what you pay. The captured amount follows your actual length, and any difference is released. If the wallet cannot cover the reserve, the request is refused with 402 insufficient_credits.

The ledger will show the reserve and then the captured amount as separate rows. That is normal, not a double charge; the refunded or released part is not spend.

How do I plan a batch of voiceovers?

Count characters once, sum them, and apply the rate. The function below does that for a list of scripts and prints the total so you can compare it with the wallet balance before you submit anything.

RATE_PER_1000 = 0.0475

def tts_cost(scripts):
    chars = sum(len(s) for s in scripts)
    return chars, round(chars / 1000 * RATE_PER_1000, 4)

scripts = ["Meet the new bottle. Zero leaks." * 10] * 40
chars, usd = tts_cost(scripts)
print(chars, "characters", usd, "USD")

Is the voice the expensive part of a video?

No. At this rate, 40 scripts of 320 characters cost about 61 cents. A single 8-second avatar clip on Standard lists at $1.47, more than twice that. Voiceover is rarely what makes a budget stretch.

What does matter is rework. A script you rewrite and re-synthesize ten times costs ten times as much, so lock the words before you generate the audio. Also, text-to-speech is separate from any speech built into a talking-avatar clip, so do not count the same words twice.

One more budgeting habit: keep scripts in a file and total them before you generate. It takes seconds and it catches the one 20,000-character paste that would otherwise reserve the maximum.

What does this not tell me?

• It does not say which voice sounds best. The catalog gives one price for the route, not a quality comparison.
• It does not cover music or sound effects, which have their own per-track pricing.
• It does not guarantee how the provider counts characters in unusual text such as emoji or markup, so keep scripts plain and check the usage ledger on your first batch.