Interview recording to transcript: how to get every word without typing it

Should you type the interview yourself, hire someone, or use an app?

Three routes, suited to different interviews.

• Type it yourself. Free, and you end up knowing the material by heart. It also takes far longer than the interview itself, because you stop, rewind and retype constantly. Fine up to about ten minutes of audio.
• Send it to a transcription service. A person listens and types. On hard audio — a panel, a noisy street, strong accents — that is still the most accurate option, but you pay per minute of audio and wait for the turnaround.
• Run it through an app. The words come back in minutes and you spend your time correcting instead of typing. For a one-to-one interview on decent audio, this is what most journalists, students and researchers do now.

Rule of thumb: the worse the audio, the more the third route costs you in editing. Two people in a quiet room need a light reread; four voices over a coffee grinder need a heavy one, or a human.

Record the interview so it transcribes cleanly

Every minute spent on the recording saves you several on the transcript, and none of this needs equipment.

• Put the microphone close. Phone flat on the table between you, screen up, nothing on top of it. Half a metre closer beats any software.
• Cut the background. Music, ventilation, an open window onto a street: transcription hears all of it as speech. Take the quietest room, even if it is uglier.
• One voice at a time. Ask for it out loud at the start. Overlapping speech is the biggest single cause of garbled lines.
• Spell the hard words. Have people spell their name, their organisation and any acronym into the recorder.
• Test twenty seconds. Record, play it back, listen — before the interview, not after.

For a remote interview, record inside the calling app when it offers that, and say on the recording that you are recording. Failing that, put a second phone next to your speaker and capture the room: worse audio, but still a transcript.

Step by step

1. Get the recording onto your phone

   Recorded on the phone? It is already there. From a dictaphone or a computer, drop the file into Google Drive, iCloud or Files, then open it from the phone. No conversion, no format check.

2. Open Zeepit, or share the audio to it

   In Zeepit, pick the file from Files or Google Drive. Or tap Share on the recording in your recorder app, WhatsApp or Drive and choose Zeepit. Same on iPhone and Android.

3. Let it transcribe, then read the summary

   The language is detected automatically. You get the full transcript, word for word, plus a summary in short, medium or long, for recordings of up to four hours.

4. Correct names, figures and jargon

   Reread against the audio. Fix proper nouns and numbers first, then add speaker labels. Budget more time for this pass than for the transcription itself.

5. Copy the text or export the document

   Copy the transcript or the summary with one tap, free. Export to PDF, Word or plain text, generated on the device, with Zeepit Pro when someone needs the file itself.

Which audio files and lengths actually work?

Most recorders already give you a usable file. iPhone voice memos come out as m4a, Android recorder apps as m4a or mp3, dictaphones as wav or mp3, and a WhatsApp or Telegram voice note as ogg or opus.

Zeepit takes all of those — mp3, m4a, wav, caf, ogg, opus, aac, flac, amr, webm — plus video files (mp4, mov, mkv, avi), using only the audio track, so a filmed interview needs no conversion. If your interview arrived as a voice message, the WhatsApp voice-notes guide covers that case step by step.

On length, one recording can run up to four hours. A whole day of fieldwork saved as a single file is better split into parts before you start; the reread becomes manageable too.

Two habits worth keeping: rename the file right after the interview (2026-08-29-lopez-clinic.m4a beats REC0043), and keep the original audio until the work is published, because you will need it to check a quote.

Verbatim or clean, and who said what: the editing pass

Decide which document you are making before you start fixing it.

• Verbatim keeps hesitations, repetitions and false starts. Research interviews, legal statements and anything where a pause is data need this.
• Clean drops the fillers and keeps the sentences — what an article, a report or a set of show notes needs.

Then make one pass, in this order:

1. Play the audio slightly faster than normal and read along, correcting as you go.
2. Fix proper nouns, job titles, figures and jargon first: that is where errors do real damage.
3. Mark the three or four passages you might quote, and listen to those again at normal speed.

Speaker labels belong to the same pass. Not every tool separates voices, and those that do stumble when people talk over each other. The low-tech fix: say the names into the recorder at the start, note the clock time of strong answers on paper, then add Q: and A: yourself while rereading.

What to do with the transcript once you have it

The raw text is rarely the deliverable.

• Read the summary first. A short, medium or long summary gives you the spine of the piece before you work through forty pages of text.
• Pull the quotes you marked into your draft while the audio is still in your ear, wording intact.
• Translate if you have to, but quote from the original language and have a native speaker check anything contentious.
• Export or copy. Copying the transcript or the summary with one tap is free in Zeepit; PDF, Word and plain-text export, generated on the device, is part of Zeepit Pro — useful when an editor or a case file needs the document itself.

Do all of it on the phone that holds the audio: fewer transfers, fewer files lost between devices. See the three-step version.

Who can hear your interview recording?

Two separate questions, and they get mixed up.

Consent. Tell the person you are recording, on the recording, before the first question. Rules differ by country and by setting — newsroom, university ethics committee, workplace — so check what applies where you work, and keep that opening line in the file.

Storage. Any tool that transcribes for you sends the audio somewhere, unless it runs entirely offline. Ask three things of whatever you use: is the file deleted after transcription, is it kept for training, and where does the transcript end up?

With Zeepit, the file goes to the servers, is transcribed and deleted right after — no archive, no reuse, no AI training on your recordings — and the transcript and summary stay on the phone. For sensitive interviews, add one habit of your own: delete the audio from the phone and from your cloud backup once the work is out.

Frequently asked questions