What transcription actually costs, and why the rates differ

Why transcription is priced per audio minute

Transcription is priced by the length of the recording, not the length of the document it produces. A sixty-minute interview costs the same whether the participants spoke quickly or slowly, and whether the finished transcript runs to twelve pages or twenty.

That unit is chosen for the buyer's benefit. You know the length of your recording before you commission anything, so you can calculate the cost yourself and approve it in advance. Per-page and per-word pricing settle the number after the work is done, at a point where you have no practical way to check it and no ability to decline.

It also removes an incentive that works against you. A provider paid by the word has a reason to keep every repetition and false start. A provider paid by the audio minute does not.

What sits behind the rate

The cost of a transcript is almost entirely time. A recording does not take an hour to transcribe because it is an hour long. It takes several times that, and the multiple grows with how difficult the audio is.

Most of that time is not typing. It is listening again. A first draft can be produced comparatively quickly. Establishing that the draft is correct means returning to the audio, passage by passage, and checking what was actually said against what was written down.

Review is the part that cannot be compressed without changing the product. A transcript that has been typed but never checked against the recording is a draft, whatever it is sold as. The rate reflects the review, which is why two quotes that look similar can describe very different work.

What moves a file between tiers

We publish three rates. AI-drafted, human-reviewed is $0.89 per audio minute, which is $53.40 per audio hour. Human from first pass is $1.19 per audio minute, or $71.40 per hour. Specialist is $1.49 per audio minute, or $89.40 per hour.

Two things separate them. The first is how the initial draft is produced: by a machine, or by a person typing from the audio. The second is whether the subject matter requires a reviewer who works in that field routinely, which is what the specialist tier provides.

What does not change is the review. Every tier receives multi-pass human review, and every tier is held to the same 99.00% accuracy target, measured by verified-word methodology. The tiers describe how a transcript is made. They do not describe how carefully it is checked.

Estimating a file before you commission it

Multiply the length of your recording by the per-minute rate of the tier you want. A ninety-minute board meeting on the human tier is ninety multiplied by $1.19, which is $107.10. The hourly figures are there if you prefer to work in hours.

The number that matters is recording length, not file size. A high-bitrate stereo file and a compressed phone recording of the same conversation cost the same. Check the duration in your player rather than the megabytes on disk.

Where a recording is difficult, or where the subject matter needs a specialist reviewer, we tell you when we quote, before any work begins. We do not revise a rate after delivery.

What a low rate usually means

Rates well below this band are common, and they are not usually a bargain. They generally describe one of two things.

The first is raw machine output. Automatic transcription is inexpensive because no person is involved. What arrives is a draft, and the work of making it correct has been transferred to you.

The second is review compressed into a single quick read. A reviewer who is not returning to the audio cannot catch the errors that matter, because the errors that matter read perfectly well on the page. They are only visible against the recording.

Neither approach is dishonest if it is described plainly. The difficulty is that both are often sold using the same words as fully reviewed work. When you compare quotes, the useful question is not what the rate is. It is how many times a person listened to the audio.