Transcribe all the interviews in a study in one batch
Upload the 10 or 20 interviews of a study together and get each one transcribed word for word as its own TXT file, ready to import into coding software, plus one merged document for reading. One payment of $5 covers up to 6 hours of interviews in total, and Unlimited covers up to 12 hours per job. A batch of up to 15 minutes is free in full.
- A separate TXT and SRT per interview, numbered in name order
- Timestamps from the start of each interview, not the batch
- Optional translation under every paragraph for interviews in another language
- No coding, themes or interpretation added by the AI
- Need speaker data per turn? The interview tool gives a CSV for each interview
From a folder of recordings to documents you can code
Qualitative fieldwork usually ends with a folder of recordings: P01 to P18, a couple of focus groups, maybe a pilot. Transcription is the bottleneck before analysis can start. Upload the folder as files or as a ZIP, and each interview comes back as its own text file, numbered in order and named after the recording, so participant codes in file names carry through.
NVivo, ATLAS.ti and MAXQDA all import plain text documents, so the TXT files can go into a project as one document per participant.
Bulk, or one interview at a time?
This page uses bulk transcription: one job, one payment, every interview. Paragraphs get a speaker label when the speaker clearly changes. If your analysis depends on who said what in every turn, the interview transcription tool fits better: it labels every paragraph with a speaker, adds a neutral summary and quotes, and gives a CSV of speaker turns with the columns File, Time, Speaker, Text and Translation. That tool takes one interview session per job, each up to 6 hours for $5.
Before you upload
- Name files with participant codes rather than real names, such as "P07 2026-03-14.m4a"
- Make sure your ethics approval or data agreement allows processing by an outside service
- Pick the spoken language if all interviews share one
- Choose a translation language if a supervisor or co-author reads another language
What to check afterwards
Read each transcript against the audio, at least wherever you plan to quote. Speech models miss words in noisy cafés, over phone lines and when people talk over each other, and they can mishear names, places and numbers. Speaker numbers are assigned per ten-minute part, so the interviewer can turn into "Speaker 2" later in a long interview; fix that with Find and replace before you download. The transcripts are drafts made by an AI speech model, not verified or certified transcripts.
How the tool works, what it costs and how your files are handled is on the Bulk Audio Transcription page; prices for every tool are on the pricing page.
Last updated
Questions and answers
Can ChatGPT transcribe all my research interviews?
It may manage a short excerpt. A study’s worth of hour-long interviews runs into upload limits and reply length, so transcripts come back shortened or partial, which is not usable for coding.
What if the study has more audio than one payment covers?
Split the interviews into batches of up to 6 hours, each $5, or use Unlimited at $19 a week, where each job holds up to 12 hours.
Will it code or summarize my interviews?
No. Bulk transcription writes out what was said and nothing more. Coding and interpretation stay with you.
Can I import the transcripts into NVivo?
Yes. The TXT files, or the Word version of the merged document, import as documents in NVivo, ATLAS.ti and MAXQDA.
Is participant data used to train models?
No. Recordings are visible only to you and are sent to an AI model provider only to be transcribed. Jobs started without an account are deleted after 7 days, and you can delete a job sooner.