Asset 20 8 2
Does AI recommend your business? Run the free check →

Join 15,000 business owners, marketers and entrepreneurs. The Sunday newsletter you'll be annoyed only arrives once a week.

Article

Do Transcription Jobs Still Exist in Today's Market

If you are skim reading
Yes, transcription jobs still exist, but the version most people picture (headphones on, typing every word from an audio file) has mostly been eaten by AI transcription engines.

Yes, transcription jobs still exist, but the version most people picture (headphones on, typing every word from an audio file) has mostly been eaten by AI transcription engines. The work that is left pays either very little (correcting AI output at speed) or reasonably well (specialist legal, court and niche foreign-language transcription where accuracy matters and someone will get sued if it is wrong).

Is it worth doing? Only if you aim at the specialist end. The US Bureau of Labor Statistics still counts medical transcription as a real occupation, with a median wage of $40,410 a year in May 2025, but it also projects that employment will fall 4 percent between 2025 and 2035. General transcription on open marketplaces is the weakest corner of all, so if you try it, treat it as a test and measure what you earn per hour from day one.

What changed, and when

I started paying attention to this around 2019, when I had a small team helping me repurpose webinar and podcast content for clients. Back then a freelance transcriptionist on a platform like Rev or GoTranscript was paid per audio minute, and a fast typist could make a modest hourly wage from it; the platforms don't publish their historical rates, so treat any old figure you see quoted online with caution. By 2023, tools like Otter.ai, Descript and OpenAI's Whisper model could turn clean, single-speaker audio into a usable draft in minutes. Accuracy varies with the recording, and no single figure covers every file, but for a clear podcast or webinar the draft was good enough that a human only needed to correct it. A job that took a person two or three hours per recorded hour became a few minutes of machine time plus a check.

The US Bureau of Labor Statistics tracks this in its Occupational Outlook Handbook for medical transcriptionists specifically. Its current page projects employment to decline 4 percent from 2025 to 2035, while still expecting about 6,200 openings a year, all of them from people retiring or moving to other jobs rather than from growth. The same page puts the median wage at $40,410 a year in May 2025, with the lowest 10 percent earning under $28,250 and the highest 10 percent over $55,950. That's employed medical work, not marketplace typing, but the direction is clear. This was already a shrinking field. AI just sped up the shrinkage.

The uncomfortable number nobody quotes accurately

Here is the bit that gets glossed over in most "is transcription a good side hustle" articles. General transcription rates on the big platforms have not just stayed flat, they have dropped in real terms. Per-minute rates for new transcribers on the big marketplaces are reported to sit well under a dollar per audio minute, and those platforms change them without much notice, so check the current rate inside your own account rather than trusting any figure quoted online. Once quality control sends back a chunk of your submitted work, a beginner's effective hourly rate can easily land under minimum wage in both the UK and the US. I have spoken to two former clients who tried this as a "quick income between contracts" in 2024 and both quit within six weeks, not because the work was hard but because the maths did not work once you counted research time, playback slowdowns for mumbled audio, and rejected submissions. If you're sizing up Rev itself, I've looked at whether Rev is legit for freelancers.

The reason this matters is that most people googling "do transcription jobs still exist" are not asking out of curiosity. They are looking for a realistic income source, often flexible, often from home. So run the numbers before you sign up. Here's a worked example with assumed figures, not any platform's published pay: say you're offered $0.25 per audio minute, and cleaning up a file takes you 4 working minutes for every audio minute. A 60-minute file pays $15 and takes 4 hours, which is $3.75 an hour. Get faster, down to 2 working minutes per audio minute, and it's still only $7.50 an hour.

To work out your own figure, log the first ten files: audio length, minutes worked including replays and research, and what you were paid after any rejections. Divide total pay by total hours. If the answer is below what you'd earn in a local part-time job, pure typing-from-audio work is a hobby that occasionally pays, not an income, unless you're extremely fast and extremely selective about which files you accept.

Where the actual work has moved

The jobs that still exist and still pay reasonably fall into three buckets, and they all look different from what "transcriptionist" meant in 2015.

  • AI editing and QA. Companies like Verbit and 3Play Media (which supply captioning and transcripts to universities, broadcasters and legal firms) now run audio through AI first and pay humans to correct the output, flag speaker changes, and check for compliance-grade accuracy. These companies don't publish a standard pay rate for editing roles, and it varies with the client and turnaround required, but the work rewards speed and judgment over raw typing.
  • Legal and court transcription. Court reporters, deposition transcribers and legal transcriptionists working with specialist firms are paid more than general transcribers, often per page rather than per minute, because the accuracy bar is far higher (a missed "not" changes the meaning of testimony) and the training barrier keeps casual competitors out.
  • Foreign-language and dialect transcription. AI models still struggle badly with strong regional accents, code-switching between languages, and low-resource languages. TranscribeMe and similar platforms pay noticeably more for these jobs precisely because AI cannot do them reliably yet. This is the one area where I would tell someone bilingual to look into it, if I were still allowed to use that word.

What makes a language job pay more is the detail around the words. You'll be asked whether the client wants full verbatim (every um, false start and repeat) or clean verbatim (tidied but faithful), to add timestamps at a set interval such as every 30 seconds, to tag the point where a speaker switches language, and to keep a glossary of names and places so they're spelled the same every time. Those are exactly the things AI gets wrong on accented or mixed-language audio, so your sample should show all four.

Everything in between, the plain English podcast transcript, the standard interview, the one-speaker webinar, is now largely a machine job with a light human check.

What the AI-editing job looks like day to day

You're handed an AI draft and the audio. You play it at normal speed with the transcript open, fix misheard names and jargon, add or correct speaker labels, check every number and timestamp, then submit against the client's style guide. Most rejections come from the small stuff: a missed speaker change, a wrong tag, or a style rule ignored. Before you apply anywhere, test yourself: take a 5-minute podcast clip, run it through a free AI transcription tool, and time how long it takes you to get it perfect. That single number tells you more about whether this work suits you than any job advert.

A real example from my own work

I used to have a virtual assistant who did light transcription for me alongside admin tasks, turning client interviews into blog drafts. In 2022 that meant her sitting with headphones on for two or three hours per hour-long recording. By late 2024 her job on the exact same task had changed completely. She runs the recording through an AI tool first, gets a rough transcript in about four minutes, then spends thirty to forty minutes cleaning it up, fixing names it mishears (it still cannot handle "Lilach" without training), adding punctuation the AI gets wrong, and restructuring rambling answers into usable copy. Same output, roughly a fifth of the time, and her value shifted from typing speed to editorial judgment. I did not cut her hours. I gave her more clients' interviews to handle in the time saved. That is the actual shape of this job now, for anyone doing it inside a business rather than on a freelance marketplace.

How to get started, if you still want to try it

If you want to test the water without wasting a month, here is the realistic path.

Work with me

Want AI doing the heavy lifting in your marketing?

I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.

  • Sign up for one AI-editing platform first, not a pure transcription one. Verbit, 3Play Media and Rev's own transcript editors all recruit for correction and QA work, and the entry bar is usually a typing test plus a short quality assessment. Expect the assessment to give you a short audio clip and a style guide, then mark you on how closely you follow it, not just on speed.
  • Do the free typing and grammar test, and time yourself, because most platforms reject you if your accuracy or speed falls below their threshold, and the rejection stage is where most beginners quietly give up.
  • Track your actual hourly rate for the first ten files, including time spent on playback, research, and any rejected work sent back for correction. Most people skip this and end up believing they earn more than they do.
  • If you speak a second language fluently, especially one with regional dialect variation (Arabic, Hindi, various African and South Asian languages), apply directly to specialist agencies for that language pair rather than general platforms, because the pay gap is significant and the AI competition is much weaker there.
  • If you want the legal or medical route, look for firms offering training and certification (some transcription companies pay for a short course as part of onboarding) rather than trying to freelance your way in cold, because clients in those fields want a paper trail of competence.

Who should not bother

If you are looking for a straightforward, low-effort income top-up, this is no longer that job, and I would rather say that plainly than let someone spend a week setting up accounts for work that pays less than stacking shelves. If you are a fast, accurate typist who is bilingual, has legal or medical sector knowledge, or enjoys the editing side of language work, there is still a living to be made here. If your plan is to transcribe clean single-speaker podcasts and webinars, you're competing directly with free AI tools that already handle that audio well, and that is not a fight worth picking.

If you do go ahead in the UK, the income counts as self-employment. HMRC's guidance on tax-free allowances says you can earn up to £1,000 a year in gross trading income under the trading allowance, and that above that you'll generally need to register for Self Assessment. Keep a simple record of every payout from day one, because marketplace payments arrive in small amounts and add up faster than you'd think.

One transcription career that holds up is live court work, explained in how to become a court reporter.

If the AI tools themselves interest you more than the typing, video is where they're moving fastest, and Veo vs Sora compares two of them.

Frequently asked questions

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

There's no official figure, because it depends on audio quality, accents, the number of speakers and how fast you type. Working transcribers commonly describe it in hours rather than minutes per audio hour. Measure your own over your first ten files and use that number, not an advert's, to judge the pay.

Is transcription a good work-from-home job for beginners?

It is accessible but low paid at entry level, often working out to under minimum wage once you count rejected submissions and playback time, so it suits people testing the water part-time rather than anyone needing a reliable primary income.

Is transcription income taxable if I do it casually?

In the UK, yes, it's trading income. HMRC's guidance says up to £1,000 a year of gross trading income is covered by the trading allowance; above that you'll generally need to register for Self Assessment. Check the GOV.UK pages for the current rules before your first tax year ends.

What transcription companies are still hiring in 2026?

Rev, GoTranscript, TranscribeMe, Verbit and 3Play Media all recruit ongoing, with the AI-editing and QA roles at Verbit and 3Play Media generally paying better than pure transcription work on the more general platforms.

Where to check the details

Run a business, a brand or a blog? I feature tools, products and expert guest posts for an audience of 15,000. See how it works

Published and maintained by the Lilach Bullock team, covering marketing, AI and business growth.
Your buyers are asking AI who to use. Does it say you?

See for free whether ChatGPT, Claude, Perplexity, Gemini and Google name you, and get the plan to become the answer.

Check my AI visibility →
Sundays only

Get the Sunday newsletter.

One email a week. AI experiments, marketing tactics, and the workflows Lilach is building right now in her own business.

Subscribe free

Let’s get your marketing running on AI.

Book a free 30-minute call

We figure out what you need, where AI fits in, and what working together would look like.

Book the call →

Or take the 30-second calculator

You’ll see the hours and the money quietly leaking out of your week, and the three workflows worth building first.

Take the calculator →

Or grab the free AI resource library

Prompt packs, templates, checklists, and swipe files. The exact tools I build for paying clients. Yours, free.

Get the library →
Keep reading

More from the blog.