How to Transcribe an Interview: A Reporter's Workflow

Every reporter has done the math at least once: a 40-minute interview takes roughly four hours to transcribe by hand if you're careful about it. Do three interviews for one story and you've lost most of a workweek to typing, before you've written a word. That arithmetic is why interview transcription was, for decades, the least glamorous and most dreaded part of the job. It doesn't have to be anymore — but the tools change the workflow, not the judgment.

This is a working reporter's approach to interview transcription: how to record so the transcript comes out clean, how to run it through a tool, and how to turn raw text into quotes you can actually publish. The goal isn't a perfect machine transcript. It's getting to accurate quotes faster, with your editorial standards intact. And whether you're doing a formal sit-down or a quick doorstep interview, the same principles hold — the format changes, the discipline doesn't.

Record like the transcript depends on it — because it does

The single biggest factor in transcript quality isn't the software. It's the audio you feed it. A great tool on bad audio gives you garbage; a decent tool on clean audio gives you something you can work with immediately. Get the recording right and you've won most of the battle.

Gear that punches above its price

You don't need a studio. You need to beat the phone-in-pocket default.

• A lavalier or external mic clipped near your subject beats a phone across a café table by a wide margin.
• A dedicated recorder (or a decent voice-memo app) recording to a lossless or high-bitrate format keeps detail the AI needs.
• Two recorders if the interview matters — a backup has saved more stories than any single piece of gear.

Managing the room

Where you record matters as much as what you record with.

• Pick the quiet corner, not the table by the espresso machine.
• Put the mic closer to your subject than to yourself — you can hear your own questions fine in your memory.
• Kill the obvious noise: fans, notifications, the clattering AC.
• At the top, have your subject state their name and title. It anchors the transcript and gives you clean attribution later.

Consent and the record

Tell your subject you're recording and get a clear yes on tape. Beyond the ethics, in many places recording without consent is illegal, and a quick recorded agreement protects you and them. This is also just good practice — the same discipline applies whether you're doing a one-on-one or moderating a group; the mechanics of a how-to-transcribe-a-focus-group session build on exactly these habits with more voices in the room.

From audio file to draft transcript

Once you've got a clean recording, the transcription step is fast and mostly hands-off.

Uploading to Transkio

Drop your audio file into Transkio through the audio-to-text tool. It runs in the browser, so there's nothing to install on a deadline — upload the MP3, WAV, or M4A and get a transcript back in a fraction of real time. Transkio supports 50+ languages, which matters if you interview sources who aren't speaking English.

A couple of practical limits to plan around: the free plan takes files up to 100 MB and recordings up to 30 minutes. A standard interview fits; a long, rambling one may need a paid plan or a split file. If you record in a high-bitrate lossless format, your files get big fast, so keep an eye on size.

AI-generated transcripts may contain errors — please review before relying on them.

Speaker labels for a two-person interview

For a straight one-on-one, you often don't need automatic speaker detection at all — you know which voice is yours and which is the source's, and the back-and-forth structure makes it obvious on the page. For anything with three or more voices, speaker detection (Elite+) helps a lot, separating speakers by voice. It's strongest when voices are distinct and people take turns, and it stumbles on cross-talk or similar-sounding speakers. Treat the labels as a draft and correct them as you review.

The review pass is the real work now

Here's the mindset shift. The tool didn't replace your transcription; it replaced the typing. The review — checking the transcript against the audio — is still yours, and for journalism it's non-negotiable.

AI transcription is good, often very good on clean audio, but it makes specific kinds of mistakes you have to catch:

• Proper nouns and names — it guesses, and it guesses wrong on anything unusual.
• Technical terms and jargon — an unfamiliar term becomes the nearest common word.
• Numbers — a misheard figure in a quote is a correction waiting to happen.
• Homophones and near-misses — small errors that change meaning.

For any quote you're going to publish, listen to that passage against the transcript. Not the whole interview — just the parts you're quoting. That targeted check is fast and it's what keeps you honest.

Turning the transcript into a story

A clean transcript is a means, not an end. The point is to write faster and quote accurately.

Pulling quotes you can stand behind

Search the transcript for your themes. Mark the strong quotes. For each one you intend to use, verify it against the audio, then decide how to handle it editorially — tightening for length, using ellipses honestly, never changing meaning. The transcript makes finding the quote instant; your standards decide what happens next.

Keeping an accurate archive

Interview transcripts are worth keeping. A searchable record of what a source actually said protects you months later when someone disputes a quote. It also makes the next story easier — you can search across old interviews for a name or a topic. The interview transcription use-case page has more on organizing this kind of archive for ongoing reporting.

When you need it in another language

If you interview a source in one language and need the quotes in another, translation is available on paid plans (Pro+), and the audio-translation tool handles the language handoff in the same upload step. Machine translation is a starting point, not a finished product — for a published quote from a non-English source, have someone fluent verify the rendering, ideally against the original audio and not just the transcript. A word choice that reads fine in English can flatten a nuance the source clearly intended. The convenience is real; the responsibility for accuracy stays with you.

Formatting for the newsroom

If your desk wants transcripts filed in a particular way, the free plan exports TXT, SRT, VTT, and a formatted Word document is available on paid plans (Pro+) if you need to hand editors something tidy with quotes pulled out.

What the tool changes, and what it doesn't

It's worth being clear-eyed about this, because overselling transcription tools does reporters a disservice. Transkio produces an AI-generated draft — fast, and accurate enough on good audio to save you those lost hours. It does not produce a certified or legal-grade transcript, and it isn't a substitute for your own verification.

What genuinely changes: the four hours of typing become fifteen minutes of upload-and-wait plus a focused review of the parts you'll actually use. You get to spend your time reporting and writing instead of scrubbing back and forth over audio.

What doesn't change: the judgment. Deciding what's a quote and what's throat-clearing, catching the misheard name before it runs, honoring what a source meant — that's the job, and no tool does it for you. Interview transcription used to eat your week. Now it's a quick step in a workflow where your editorial standards still do the heavy lifting. That's exactly how it should be.