How do you get a transcript without using AI?

You get a transcript without AI in one of two ways. You can listen to the recording and type it yourself, with playback software that lets you pause and replay easily. Or you can pay a human transcription service, which uses people rather than machines to produce the text. The first route is free but slow. The second costs money and returns a finished file in hours.

Route one: listen and type it yourself

Doing it by hand needs little more than the recording and a word processor. The slow part is the replay. A Cornell interview project guide tells interviewers to expect 4 to 6 hours of transcribing for each hour of recorded interview ¹.

Playback software makes this faster. Express Scribe, from NCH Software, can be run from keyboard hot keys or a foot pedal ². Its free version supports common audio formats such as wav and mp3, and the page says the free version does not expire ³. A foot pedal lets you start and stop with your feet while your hands stay on the keyboard, which is the setup typists have long used ⁴.

Typing speed sets the total time. A one-hour recording at 150 words per minute holds about 9,000 words, which takes about 175 minutes of pure typing at the 51.56 words per minute average from a large online study ⁵. You will pause and replay, so the real total is longer. The typing speed guide covers how to type faster, which shortens this route most.

Route two: pay a human transcription service

A human service takes your recording and returns a text file. Rev's human service charges $1.99 per minute, with a stated accuracy of 99 percent or better and delivery in 12 hours or less ⁶. For a one-hour recording, the list price is about $119 ⁷. Services differ in price and turnaround, so compare the figures on each provider's current page.

This route is the one to choose when you cannot spare the hours. Court and other legal records can have their own rules about who may prepare them, so check those before you hire a general service. The guide on human versus AI transcription explains the rule.

What "without AI" means

Many people use "AI" to mean any machine speech recognition, and others mean only the newer generative models. Dictation and voice typing use speech recognition, so if you want no machine-written text at all, the listen-and-type route and the human service are the ones that fit. The question of whether transcription software counts as generative AI depends on definitions, which the answer on whether transcription software is generative AI explains.

Dictation is a different job. When you speak your own words, you are writing rather than transcribing a recording. Yap, which we make, is an on-device dictation app for Mac and Windows, and it is built for that work, not for transcribing a finished recording. You can download it here.

When the by-hand route is enough

• Short recordings of a few minutes, where the typing time is small.
• Private notes where you can accept some gaps, because you can fill them from memory.
• Practice, such as learning a language or a shorthand system, where typing the words is the point.

For longer or important recordings, the time adds up fast. A one-hour interview typed by hand can take four to six hours ¹, most of a working day, and a service may be worth its price when your own hours are scarce.

What to do

1. Choose by time and budget. If you have the hours, type it yourself with a playback tool. If you do not, pay a service.
2. Set up the playback first. Install a playback program, set a comfortable speed, and try a foot pedal or hot keys before you start.
3. Type in short sections. Transcribe a few minutes, check it, and move on. Long sessions produce more errors.
4. Check names and numbers. Those are the easiest things to get wrong when typing quickly.

For the other routes, see how much transcription costs per minute and how accurate AI transcription is. The guide on human versus AI transcription compares the routes on accuracy, cost and speed, and our overview of transcription covers the subject as a whole. A playback control that you run with your feet is a transcription foot pedal.