Preview Sume TTS sentence ids, lengths and job cost before you submit

Before you accept a script and generate, you can predict three things offline: the sentence ids the source API will assign, the character count of each, and what each job will cost. The cost rule is Sume TTS's list price of $0.0475 per 1,000 characters, rounded up to the next cent per job, with a one-cent minimum and a 20,000-character cap per request.

Integer math avoids float surprises: cents = ceil(characters x 19 / 4000), which is the same as 4.75 cents per 1,000.

The script

It uses the same sentence rule as the source API (cut after ., !, ?), then packs consecutive sentences into jobs under the cap. It does not call the API, so it needs no key.

import re

def cents(chars):
    return max(1, -(-chars * 19 // 4000))

def plan(text, cap=20000):
    parts = [p for p in re.findall(r"[^.!?]*[.!?]|[^.!?]+\Z", text) if p.strip()]
    jobs, cur = [], ""
    for p in parts:
        if cur and len((cur + p).strip()) > cap:
            jobs.append(cur.strip())
            cur = ""
        cur += p
    if cur.strip():
        jobs.append(cur.strip())
    return jobs

script = "Welcome back. Today we cover three tips! Ready? " * 400
jobs = plan(script)
for i, j in enumerate(jobs):
    print(i, len(j), cents(len(j)))
print("total cents", sum(cents(len(j)) for j in jobs))

What it prints

The sample script is 19,200 characters before trimming. Cost depends on how the packing lands, so run it on your own text. The per-job rounding is the only place the total can exceed a flat per-character estimate.

Caveats

• The server also adds a newline when a selection crosses a block boundary; keep a small margin under 20,000.
• This predicts the price of the literal characters. It does not know your voice or model; the table price is for Sume TTS 1.0 at the list rate.
• A sentence longer than the cap needs a rewrite; the splitter above would put it in a job of its own and the API would refuse it.