How to Transcribe an Interview (With Speaker Labels and Timestamps)
A practical 4-step workflow for researchers, journalists, and podcasters: verbatim vs. clean read, speaker diarization, word-synced checking, and exports.
Updated October 14, 2026Michael NguyenTested October 2026
Quick answer
Typing a one-hour interview by hand takes 4 to 6 hours; running it through an automatic transcription tool with speaker diarization and word-synced playback cuts the job to a few minutes of quote checking. Audo AI's Transcribe tool (/transcribe) converts audio and video interviews into text across 60 languages, labels up to 8 speakers (Speaker 1, Speaker 2), and exports TXT, SRT, VTT, or word-level JSON—free for the first 2 minutes with no sign-up, and $1.20 per hour ($0.02/minute) for full recordings up to 10 hours.
| Method | Time for 1 hr of audio | Typical cost | Speaker labels & timestamps | Export formats |
|---|---|---|---|---|
| Manual typing | 4 to 6 hours | Your time (or $60–$150/hr human service) | Manual only | Text document |
| Word processor dictation / upload | ~5–10 minutes + manual cleanup | Bundled with office suite (often capped at 300 min/month) | Basic speaker turns, sentence timestamps | DOCX / copy-paste (no SRT, VTT, or word JSON) |
| Audo AI Transcribe | Under 1 minute for a 15-min clip (2-min preview in <10s) | First 2 min free (no sign-up), then $1.20/hr ($0.02/min), no subscription | Up to 8 speakers + every word's exact start and end time | TXT, SRT captions, VTT (with speaker tags), and word-level JSON |
Listen and read along
Press play to see who says what.
Transcript
- Speaker 1
Welcome back to the show! So, you recorded your whole first album in your kitchen?
- Speaker 2
I did, yeah. Honestly, the kitchen had the best sound in the flat.
- Speaker 1
Wait, really? Better than a studio?
- Speaker 2
Well, it was free. We hung blankets over the windows, waited for the street to go quiet, and pressed record.
- Speaker 1
I love that.
Voices in these examples are AI-generated.
Step 1: Choose verbatim or clean read before you transcribe
Before you upload your recording, decide how the transcript will be used:
- Clean read (Keep filler words off). Leaves out 'um', 'uh', and false starts so quotes read smoothly in journalism articles, podcast show notes, and blog posts.
- Verbatim (Keep filler words on). Preserves every 'um', 'uh', hesitation, and repeated word—essential for qualitative academic research, legal records, linguistic analysis, or when you plan to cut filler words from the audio itself.
Step 2: Turn on speaker labels (diarization)
An unlabeled wall of text forces you to listen back to every paragraph just to remember whether the interviewer or the subject was speaking. With 'Label speakers' switched on (the default on Audo's Transcribe page), Audo separates turns into Speaker 1, Speaker 2, and so on in the order each person first speaks, for up to 8 speakers across 60 languages.
If your interview was recorded in a loud restaurant or the guest was sitting far from the microphone, running Remove noise (/noise-removal) or Balance audio (/audio-leveler) first helps keep quiet side remarks from being missed.
Step 3: Check names, acronyms, and key quotes with word-synced playback
Even the best speech recognition can misspell an unfamiliar surname, company name, or medical term. Instead of listening to a 60-minute interview from start to finish, use Audo's interactive transcript player (or click 'Open in editor'):
As the recording plays, each word lights up in its speaker's color. Clicking any word jumps the audio to that exact fraction of a second, so you can verify direct quotes and proper nouns in two or three minutes.
Step 4: Export the format your workflow needs
One transcription run on Audo gives you four outputs with no extra charge:
- Plain text (.txt). One paragraph per speaker turn ('Speaker 1: …'), ready to paste into research coding software (NVivo, Atlas.ti), Google Docs, or Notion, plus a simple find-and-replace to swap 'Speaker 1' for the person's real name.
- SRT captions (.srt). Numbered, timed subtitle blocks with no speaker prefixes and no caption spanning two speakers—ready for YouTube, Premiere, Final Cut, or DaVinci Resolve.
- WebVTT (.vtt). Timed web captions with <v Speaker 1> voice spans for HTML5 video players.
- Word timings (.json). Every sentence and word with its exact start time, end time, and speaker number for custom scripts and AI pipelines.
When another transcription tool is the better choice
Pick the tool that matches how your team works after the transcript is generated:
- Certified legal stenographers or human transcription. For courtroom proceedings, depositions, or sworn legal filings that require a certified human court reporter.
- Live meeting bots (Otter, Fireflies, Granola). If you want an automated bot to join your live Zoom or Google Meet calendar events and email a summary afterward. Audo AI works on recorded audio and video files rather than joining live calls.
- Multi-user collaborative doc editing (Descript). If a team of producers needs to leave comments and edit a shared multitrack project simultaneously inside a subscription workspace.
Or transcribe your interview in one step
Drop an audio or video recording into Audo's Transcribe tool to get speaker-labeled text, SRT captions, and word timings in 60 languages. Free to try. No sign-up.
Questions
How long does it take to transcribe a 1-hour interview?
Typing a 1-hour interview by hand takes 4 to 6 hours. On Audo AI, a 2-minute preview finishes in under 10 seconds, and full interviews typically finish in a couple of minutes, with word-level timestamps and speaker labels included.
How do I replace 'Speaker 1' and 'Speaker 2' with real names?
Automatic diarization separates voices by sound rather than guessing names, so it labels them Speaker 1, Speaker 2, and so on in the order they first speak. Download the TXT or VTT file and run a single Find & Replace for 'Speaker 1:' with your interviewer's name.
Should I clean up background noise before transcribing an interview?
If the recording only has mild room tone, transcribe it directly. If a speaker is buried under loud café noise, wind, or is much quieter than the host, running Remove noise (/noise-removal) or Balance audio (/audio-leveler) first improves accuracy on quiet phrases.
Can I transcribe a video interview (MP4 or MOV) to text?
Yes. Drop any MP4, MOV, MKV, WebM, or AVI video into Audo's Transcribe tool (/transcribe) and you get the full text transcript plus SRT and VTT captions timed to your video.
How much does it cost to transcribe a full interview on Audo?
The first 2 minutes of any recording are free to try, 3 times a day, with no sign-up. Full files cost $1.20 per hour ($0.02 per minute) from one-time credit packs starting at $5, with no subscription.