**Interview Transcripts for UX and Research Teams | SozAI**
Source: https://sozai.app/business/research-teams/

Research teams

# Interview transcripts for UX and qualitative research teams

Upload the session or record the call without a bot. Rename speakers to P1 and P2, export for your coding tool, and give the ethics board a list of everyone who touches the audio.

[Transcribe a research interview free](https://web.sozai.app/)

## The short answer

SozAI turns research interviews and usability sessions into transcripts with speakers told apart by voice and a timestamp on every line, in 99 languages. You rename speakers to participant codes on the web or in the phone app, then export VTT for coding tools that sync text with audio, TXT with timecodes and speakers, or JSON with times and a speaker for each segment. Files are stored in Germany, speech recognition runs in the US, and every sub-processor is named on one page you can attach to an ethics application. A research group buys seats at $12.99 a month or $99.99 a year each; there is no shared project repository, so each researcher keeps their own library.

## From session to coded transcript

1. Get consent for the recording and for the way it will be processed, in writing, before the session. A paragraph you can adapt is further down.
2. Record. In a lab or a café, SozAI on a phone on the table. A remote session in Meet, Zoom or Teams open in Chrome or Edge, from web.sozai.app with the audio source "Both". A session in a desktop call app, in the Mac app's Meeting mode.
3. Or upload existing recordings, several at a time, up to 500 MB each.
4. Rename the speakers to your codes (P1, P2, Moderator) before you export, so no real name leaves SozAI. Renaming works on the web and in the phone app.
5. Correct the text in the phone app if your protocol needs a clean transcript. Corrections are saved and carried into every later export.
6. Export in the format your analysis tool reads (table below) and import it there with the audio file.
7. Delete the recording in SozAI when your data management plan says so.

![SozAI web app: the upload dialog with four interview files ready to be transcribed](https://sozai.app/wp-content/themes/sozai/assets/images/screens/en/web-upload-multiple-files.webp)

## Which export for which job

| You need | Export | From | Note |
| --- | --- | --- | --- |
| A coding tool that links text to the audio and reads speakers | VTT | Phone app, Mac, iPad | Carries timestamps and the speaker codes you set |
| A plain document to code line by line | TXT with timecodes and speakers switched on | Web | Each paragraph reads like "[04:12] P3: text" |
| Your own spreadsheet or script | JSON | Phone app, Mac, iPad | Each segment has its start and end in seconds, its text and its speaker |
| A highlight clip with subtitles | SRT or VTT | Apps and web | SRT never carries speaker names |
| A read-only copy for a stakeholder | PDF | Phone app, Mac, iPad | Keeps speaker names and timestamps |

There is no Word export. Step-by-step imports into NVivo, ATLAS.ti and MAXQDA, with the formats each one accepts, are in our guide to [qualitative research transcription](https://sozai.app/interview-transcript-for-research/).

## Usability sessions record sound, not the screen

SozAI records audio only. For a remote usability test, run your usual screen recorder for the video and let SozAI make the transcript, or upload the recording your call tool saved. When you record the browser tab, the participant sees nothing new in the call: there's no bot and no extra name in the participant list. That makes it all the more important that the consent covers the recording itself.

Think-aloud sessions are where timestamps pay off. A participant who says "wait, where's the button" at 12:41 can be matched to the frame in your video without watching the whole hour again.

## What to tell participants and the ethics board

Review boards ask where the audio goes and who processes it. Then they ask when it's gone. These are the answers for SozAI as of October 2026.

- Files and recordings are stored on our servers in Germany (Hetzner).
- Speech recognition for files and recordings is done by AssemblyAI in the United States, so the audio leaves the EU for that step.
- Summaries and the other AI features send transcript text to AI model providers. Who they are, and when each one is used, is on the [sub-processor list](https://sozai.app/subprocessors/).
- Our own code never sends recordings into model training. Each partner works under its own terms, linked from that list.
- The Mac app's Meeting mode transcribes on the Mac itself: the audio is not uploaded, and only the text is saved to the account.

A consent paragraph you can adapt (check it against the sub-processor list on the day you submit, and against your institution's template):

> The interview will be audio-recorded and transcribed with SozAI. The recording is stored on servers in Germany, and the speech recognition is carried out by a service provider in the United States. Names in the transcript will be replaced with participant codes. The recording will be deleted from the service by [date], at the end of the study.

We don't give legal advice, and recording rules differ between countries. Your institution's ethics office has the final word on what the form must say.

## Deleting data at the end of a study

Deletion is in the researcher's hands. Anyone can delete a single recording, or their whole account, in the app or on the web. That removes the audio and everything made from it from our servers.

Three things to put in your data management plan, because they are true today. Encrypted database backups keep deleted data for up to 12 months. Deleting in SozAI does not remove the speech recognition partner's copy. And transcripts are never deleted automatically: audio can be (each account chooses 30 days or right after transcription), but the transcript stays until someone deletes it, so set a calendar reminder for the date in your consent form.

The full picture, including what a deletion leaves behind, is on the [security page](https://sozai.app/security/).

## Focus groups and many voices

Speaker separation works best when people take turns. With six or more participants, short interjections ("yes", "same here") often get grouped with a louder neighbor. We saw the same in a public board meeting we processed, where brief motions and votes were merged into two main voices; the example is on [SozAI for teams](https://sozai.app/business/). For focus groups, plan a pass where you check every speaker change against the audio before coding.

Languages: SozAI detects the language of each recording itself. Our [published accuracy test](https://sozai.app/sozai-accuracy-test-10-languages/) found 3.6% word errors in English and 3.0% in German on short read sentences, but 19.6% in Arabic. Interviews that switch language mid-answer have not been measured.

## A research group on team seats

The PI or a lab manager buys seats on web.sozai.app and invites each researcher by email; university groups can choose the Education plan at the same price. Each seat includes up to 700 hours of transcription. The person who buys does not see anyone's recordings. That's good for participant confidentiality, but it also means SozAI is not a shared research repository: interviews reach the project's analysis tool through exports. Payment, invoices with your institution's tax ID and VAT are described in [how to buy for a team](https://sozai.app/business/buying/).

## When SozAI is not the right tool

- You need a shared repository where the whole team tags and searches every session. Use a research repository tool and import SozAI's exports into it.
- Your protocol forbids any audio leaving the EU, and the sessions aren't on a Mac. Uploaded audio goes to a US speech recognition provider.
- You need verbatim transcripts with pauses, overlaps and non-verbal sounds marked by convention. SozAI writes clean text; mark those by hand or use a human service.
- Participants are children. Our privacy policy says the service is not intended for children under 13.

Comparing several tools for the group? Start with [how to choose transcription software](https://sozai.app/business/choose-transcription-software/). Interview work outside research is on [SozAI for hiring teams](https://sozai.app/business/hiring/) and [SozAI for newsrooms](https://sozai.app/business/newsrooms/).

Updated October 27, 2026 By [Merey Tleugazin](https://sozai.app/about/merey/)

Answers

## Frequently asked questions

Can I use SozAI for qualitative research interviews?

Yes. SozAI transcribes interviews in 99 languages with speakers and timestamps, and you can rename speakers to participant codes before exporting for an analysis tool. Check your ethics approval covers the sub-processors listed on our sub-processor page.

Where is research audio stored when I upload it to SozAI?

On our servers in Germany, at Hetzner. Speech recognition for uploaded audio runs at AssemblyAI in the United States, and every company that processes data is named on the sub-processor list.

Does SozAI export transcripts for NVivo, ATLAS.ti or MAXQDA?

Yes, through standard formats. ATLAS.ti and MAXQDA read SozAI's VTT with timestamps and speakers; for NVivo, use TXT with timecodes and speakers, or JSON rearranged into a table. There is no Word export.

Does deleting a recording in SozAI delete it everywhere?

Not everywhere. Deleting removes the audio and transcript from our servers, but encrypted backups keep data up to 12 months, and the speech recognition partner's copy is not removed.

Can a research group share one SozAI workspace?

No. Each researcher has a private library; the person who buys the seats manages them but can't see recordings. Share transcripts by exporting them into your project's analysis tool.

Can SozAI tell apart speakers in a focus group?

Yes, by voice, but check it. With six or more people, short interjections are often grouped with another speaker, so review every speaker change against the audio before you code.

Does SozAI record the screen during a usability test?

No. SozAI records sound only. Use your screen recorder for the video and SozAI for the transcript; matching timestamps link the two.

Turn your next recording into text

Record, upload a file or paste a link. You get the transcript, speakers and a summary back.

 [Transcribe a research interview free](https://web.sozai.app/)

Free on iOS and Android. No account required for YouTube transcripts.


---
This is the markdown twin of https://sozai.app/business/research-teams/ — the same content, without the markup.
Published by SozAI (https://sozai.app). Reuse and quotation are allowed with attribution and a link back.
Machine-readable index: https://sozai.app/llms.txt · data API: https://sozai.app/api/