Year-end donor voice messages for 2,000 names: $40 vs $100

Splicing a name clip into a tier message costs $0.02 per donor on Sume, $40 for 2,000 donors, against $100 for a full 900-character message each.

4 min readSume
All posts

A nonprofit that wants a personal voice thank-you for 2,000 donors can do it for $40.30 on Sume by recording one body message per donor tier and splicing in a short name clip, against $100.00 if every donor gets a full 900-character message generated on its own. GivingTuesday is Tuesday, December 1, 2026, 59 days from today, and the thank-you has to be ready before the gifts arrive.

The saving comes from how Sume bills text to speech: $0.0475 per 1,000 characters, but each job is rounded up to a whole cent. A 900-character message is $0.04275 and bills as $0.05. A 25-character name greeting is about $0.0012 and bills as $0.01. The splice avoids paying the 900-character price 2,000 times.

The two plans

Plan A generates the whole message per donor: 2,000 jobs at $0.05 is $100.00. Plan B generates six bodies once (three giving tiers in two languages, $0.05 each, $0.30), then for each donor generates a name greeting ($0.01) and joins greeting and body with timeline audio concat ($0.01), $0.02 per donor or $40.00 for the list.

Concat is sample-domain with no silence at the seam and no re-synthesis, so it is cheap and fast, but the voice, speed and loudness of the greeting must match the body: use the same voice id, language and speed in every job, and listen to ten joins from different names before you run the list. Names are the risk. Pass a pronunciation dictionary id for names your audience will notice, and keep a hold-out list of donors whose names you recorded by hand.

Donor message cost for 2,000 donors (read 2026-10-03)
PlanJobsCalculationTotal
A: full message per donor2,0002,000 x $0.05$100.00
B: 6 bodies once66 x $0.05$0.30
B: name greeting per donor2,0002,000 x $0.01$20.00
B: concat per donor2,0002,000 x $0.01$20.00
B total$40.30

Care points

A synthetic voice reading a donor's name is still a synthetic voice. Say so in the message or the email around it where your regulator or platform expects it, and never use a voice clone of a real person, such as your executive director, without their written consent. Keep the message honest about what the gift did; do not generate statements about impact figures you have not verified.

The concat job accepts up to 20 parts, so a greeting, a tier body and a closing line fit in one job. Each concat takes Sume-hosted audio only, which means the greeting and body outputs are fine inputs as they are, but any audio recorded elsewhere must be imported first.

Send an idempotency key derived from the donor id and the message version. If a batch fails halfway, rerunning the same keys returns the finished jobs instead of billing them again, and changing the message version produces new keys on purpose.

Sources

Related posts

More in Pricing

All Pricing posts

Written by Sume