Broadcast · Stations & podcast networks

Two accounts here already behave like radio stations. This is what that costs.

Not a brochure. We profiled every account on this platform and one group stood apart: hour-long captures, named by day and time slot, filed one after another. Below is the measured shape of that workload, what it costs at our actual prices, and how a station or a network plugs into it.

Talk about a station workload See the real prices

The pattern we measured

On 2026-08-03 we profiled every account on the platform by file metadata alone. The heaviest per-account load on the whole platform was not a media company, an agency, or an enterprise. It was two accounts submitting broadcast hours.

67.7 h
From two accounts

Two accounts, classified as broadcast/radio, submitted 67.7 hours of audio between them across the profiling window.

source: internal account profile, 2026-08-03
33.8 h
Each, on average

Per-account load for the broadcast group. The next-heaviest group — meetings, interviews and shows, across 367 accounts — averages 1.42 hours each.

33.8 h vs 1.42 h per account
≥20 / wk
Submitted unattended

The heavier of the two runs at bulk cadence — twenty or more jobs a week, week after week. Only three accounts anywhere on the platform do that. One is a station.

cadence class: bulk / automated

What the filenames say

We classify workloads from filenames, and station filenames are unmistakable. They carry a day of the week and a time slot, and they repeat:

STATION_Friday_10am.mp3 STATION_Wednesday_1pm-2pm.mp3

Hour after hour, day after day. Nobody sits and does that. That is a capture pipeline writing files off the logger and something downstream needing them as text — for a compliance log, for a searchable archive, or to turn the talk breaks into things you can publish. It is the one workload on this platform that behaves like infrastructure rather than like a person trying a tool.

Cadence predicts everything; volume does not

The same profile split the base by rhythm rather than size, and the split is stark. Accounts with a repeating weekly cadence are a rounding error in count and carry a disproportionate share of real work:

Bulk / automated20 or more jobs a week
3 accounts · 72.1 hours
Weekly workflow3 or more jobs a week
24 accounts · 56.6 hours
Trialtwo jobs or fewer, ever
the overwhelming majority · no rhythm at all

Twenty-seven accounts on the whole platform have an actual rhythm. If you run a station or a network, you are structurally in that twenty-seven the day you start — your audio arrives on a schedule whether anyone is thinking about transcription or not. Everything below is priced for that, not for someone transcribing one file and leaving.

What it costs

Three packs, all one-time purchases of minutes, all listed on /pricing. Credits do not expire, so an unused month is not a lost month.

$4
300 minutes · 5 hours

Enough to run a week of one daily captured hour and check the transcripts against your own ears before committing anything.

$0.013 / minuteone-time, non-expiring
$12
2,000 minutes · 33 h 20 m

One captured hour a day is about 1,800 minutes a month, so a month of daily capture fits inside this pack. It also covers the 33.8-hour average we measured for a station account.

$0.006 / minuteone-time, non-expiring
$24
5,000 minutes · 83 h 20 m

Two captured hours a day is about 3,600 minutes a month, which fits here with room for a breaking-news week or an election period on top.

$0.005 / minuteone-time, non-expiring

No free tier. There is a one-time 60-minute bonus when you create an account — enough for a real broadcast hour, which is the only test worth running. After that it is paid. If your volume sits well above the $24 pack, that is the conversation to have with us directly rather than buying packs in a loop.

The workflow, concretely

  1. Your capture keeps doing what it does. We do not replace your logger, your recorder, or your retention system. Whipscribe takes the files those produce and makes them searchable text.
  2. Submit without a person in the loop. Hold an API key and post the recording, or post a media URL and let us fetch it. A capture pipeline should talk to an endpoint, not to someone clicking upload thirty times a day.
  3. Get it back timestamped and speaker-labelled. Diarisation and word-level timestamps are request options, not upsells. There is no fixed length cap — a multi-hour logger file is the intended shape, not an edge case.
  4. Export into whatever consumes it. Plain text, JSON, SRT, VTT or DOCX from the same job. SRT and VTT go straight to captions; JSON with word timings is what you index if you are building search over your own archive.
  5. Keep or delete on your schedule. Retention is yours to set. Nothing is forwarded to a third-party transcription or AI service, and nothing is trained on your audio — we run open-source Whisper on machines we operate.

If you are a podcast network, not a station

Same shape, different label. A network with a dozen shows publishing weekly is a recurring capture workload with a schedule attached — the cadence class the numbers above describe. We already run a transcript and episode-summary pipeline over public-domain and Creative-Commons shows at /podcasts, so per-episode transcripts, timestamps and summaries at feed scale are a path that already exists here rather than something we would be building for you from scratch.

Being straight about the evidence: the measured figures on this page come from radio-shaped accounts. We have not measured a podcast network on this platform, and we are not going to imply we have. What we are claiming is that the workload shape is the same one; whether that holds for your network is something a pilot answers and a landing page does not.

Questions

What does it cost to transcribe one captured hour a day?

About 1,800 minutes a month, which fits inside the $12 pack of 2,000 minutes. Two hours a day is about 3,600 minutes and fits inside the $24 pack of 5,000.

Is there a length cap on a single recording?

No fixed cap. Multi-hour as-broadcast logger recordings are what the long-form path is for, and timestamps run the full length of the file.

Can a capture pipeline submit unattended?

Yes — hold an API key, post the file or a media URL, poll for the job, pull the result back as txt, json, srt, vtt or docx.

Is there a free tier?

No. A one-time 60-minute signup bonus, then paid. Credits do not expire.

Where does our audio go?

To our own machines. Open-source Whisper on infrastructure we operate — not forwarded to a third-party transcription or AI service, and not used to train anything.

Will you name us as a customer?

Not unless you ask us to in writing. There are no station logos on this page for that exact reason.

How the numbers on this page were produced. An internal profile of every account on the platform, run 2026-08-03, derived from file metadata only — filenames, durations, cadence, submission source, file extension, country and email domain. Transcript content was never read. Roughly 30% of accounts stay unclassified because their filenames are opaque, and we would rather say so than round it away. The classifications are group-level reads, and no station is identified here.

Not legal advice. Whipscribe makes recordings searchable; it complements your recording and retention obligations and does not discharge them. Confirm what your licence conditions and applicable code require.

Bring us an hour of your actual output

Tell us which station or network, roughly how many hours a week you air, and what you need the text for. A real person on a small team reads it and replies within one business day — no BDR queue, no discovery call before you get an answer about price.

contact@neugence.ai Pilots, DPAs & volume pricing

CC your procurement or compliance lead on the first email and we will run the whole review on one chain.