47 meeting recordings of 52 minutes: transcription cost on Sume

Transcribing 47 recordings of 52 minutes costs $24.44 on Sume STT at $0.01 per audio minute, as 282 requests of up to 10 minutes each.

5 min readSume
All posts

Transcribing 47 meeting recordings of 52 minutes each costs $24.44 on Sume STT: 47 x 52 = 2,444 audio minutes at $0.01 per minute. Because a request accepts at most 10 minutes of audio, each recording is split into 6 requests (five of 10 minutes and one of 2), so the batch is 282 requests.

The numbers

The catalog prices Sume STT 1.0 at $0.01 per audio minute and states a maximum of 10 minutes per request. If you omit duration_seconds the request reserves 1 minute, so pass the real length of each chunk.

A 52-minute recording is five 10-minute chunks and one 2-minute chunk: 6 requests, 52 minutes, $0.52. For 47 recordings, 47 x $0.52 = $24.44 and 47 x 6 = 282 requests.

Sume STT 1.0 at $0.01 per audio minute, as of 2026-10-09
Recording length (min)Requests of up to 10 minCost per recording47 recordings
10110 x $0.01 = $0.10$4.70
25325 x $0.01 = $0.25$11.75
52652 x $0.01 = $0.52$24.44
90990 x $0.01 = $0.90$42.30

Splitting without losing words

Cutting at fixed 10-minute marks can slice a sentence in half. Cut at silence, or overlap chunks by a few seconds and drop the duplicated words when you stitch the text. Overlap costs a little: 4 seconds of overlap on each of five joins is 20 seconds, or about $0.003 per recording.

Sume's Timeline and audio tools can split audio into parts; their per-job prices are small ($0.01 per concat or split job in the catalog), so splitting 47 recordings is another $0.47 if each is one job.

Budget framing

At $0.52 per recording, an hour-long weekly standup run for a year (52 recordings of 60 minutes) is 3,120 minutes, or $31.20. The cost of transcription is usually small next to the cost of reading the transcript, so the practical question is accuracy and turnaround, not price.

Remember the 1 cent minimum per request: a 20-second chunk still costs a cent, so very short chunks cost more per minute than long ones. Keep chunks close to 10 minutes.

  • Per-request maximum: 10 minutes
  • Minimum per request: 1 cent
  • Reservation default: 1 minute

A second scenario

Express it per hour of audio. At $0.01 per minute, an hour of recording costs $0.60. The 2,444 minutes in this batch are 40.7 hours, and 40.7 x $0.60 = $24.44. If next quarter you record 150 hours, the transcription line is $90.00. Use that rate to compare with any other quote, as long as the other quote counts the same unit.

Running it

Submit each chunk as a job, store the job id with the recording and chunk index, and poll the job result. Use an Idempotency-Key per chunk so a retry cannot double-bill. The Jobs and results page explains the polling endpoints.

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.

Sources

Related posts

More in Pricing

All Pricing posts

Written by Sume