How long does it take to transcribe an hour of audio?

So, how long does it take to transcribe an hour of audio? If you type it yourself, plan for most of a working day. If you use software, the transcript itself takes minutes, and your time goes into checking it instead. This guide explains where the hours go, what makes a file faster or slower, and how to estimate your own recording.

How long does it take to transcribe an hour of audio by hand?

The figure you'll see most often is a 4:1 ratio: about four hours of work for every hour of recording. Treat it as a rule of thumb from people who do the work, not the result of a study. It assumes clear audio, one or two speakers who take turns, and a typist who is reasonably fast.

The ratio is high because typing is slower than speech. People talk much faster than most of us can type, so you can't keep up in one pass. In practice you play a few seconds, pause, type, rewind, and play again. Every unclear word adds another rewind.

On top of the typing, most people do a second pass: listening through once more to catch mistakes, fix spellings and add speaker names. That review is part of the four hours, not extra to it, and skipping it is where errors slip into the final text.

What makes it faster or slower

The same 60 minutes can take three hours or eight. These are the factors that move the number most.

Audio quality

This is the biggest single factor. A recording made with the phone close to the speaker is far easier to follow than one made from across a room. Background music, traffic and air conditioning all cost you rewinds.

Number of speakers

Each extra voice adds work. You have to work out who is talking, label them, and untangle moments when two people speak at once. A focus group with six people can take far longer per hour than a one-on-one interview.

Verbatim or clean

A clean verbatim transcript drops filler words, false starts and repeated words. Full verbatim keeps every "um", "you know" and stutter, which some researchers and legal uses need. Full verbatim is slower because you can't smooth anything over.

Human transcription services: turnaround times

If you pay a person or an agency, the question changes from "how long will I spend?" to "how long will I wait?" Human services usually sell several turnaround options, and faster turnaround costs more.

Standard turnaround for an hour of audio is often counted in days. Rush options can bring it down to the same or next day. Exact times and prices vary by provider, so check the current terms before you commit. For a side-by-side look at costs, see pay per file or subscription and the free transcription cost calculator.

A human service saves your time but not your wait. If your deadline is tomorrow morning, that matters.

AI transcription: minutes, plus your review

Speech recognition software processes audio far faster than a person can type. An hour-long file is usually ready in minutes rather than hours. The trade-off is that the first draft can still have mistakes, especially with names, jargon and crosstalk.

So the real time cost of AI transcription is the review. For a clear recording, reading through and fixing a handful of names may take a fraction of the recording's length. For a noisy file, you may need to listen to the tricky parts again.

With Allnote on the go, you upload the file, pick the language (or let it detect the language), and get a transcript with speaker labels and timestamps. You can then:

1. Read the free preview: the first 5 minutes, or half the file if it's under 10 minutes.
2. Pay for the file, or use credits from a pack. There's no account and no subscription.
3. Fix text in the browser and use find and replace to correct a name everywhere at once.
4. Rename "Speaker 1" and "Speaker 2" to real names.
5. Download Word, TXT, SRT or VTT, or copy the text.

Worked example: a 60-minute interview

Say you're a graduate student with a 60-minute interview recorded on a phone. Two speakers, a quiet office, mostly clear audio. Here's how the three routes compare.

If you value your time, the math gets clear fast. The U.S. Bureau of Labor Statistics lists the median pay for medical transcriptionists at $19.43 an hour (May 2025). At that rate, four hours of typing is worth about $78 of someone's time, before any rewinds for difficult audio.

A study with 20 interviews multiplies everything. At four hours each, that's 80 hours of typing, or two full work weeks. That's why many researchers type one or two interviews by hand to learn the material, then use software for the rest. See research interview transcription for that workflow.

How to cut the time, whichever route you choose

• Record well. Put the phone or mic close to the speaker and away from fans and traffic. A good recording saves more time than any tool.
• Trim what you don't need. Cut silence, setup chatter or a long break with the free audio cutter. Nothing is uploaded.
• Split long files. A very long recording is easier to check in parts. The audio splitter breaks it into pieces in your browser.
• Decide on verbatim first. Choose clean or full verbatim before you start, so you don't redo work.
• Keep a name list. Write down names, places and terms as you hear them, then fix them in one go.
• Use hotkeys if typing. Pausing and rewinding from the keyboard is much faster than using the mouse.

What it costs

With Allnote on the go, you pay for the minutes you upload: 1 credit is 1 minute of audio, and leftover credits don't expire. There's nothing to subscribe to.

To estimate a whole project, try the free transcription cost calculator.

Questions

Written with AI help.