Interview transcription
Interview transcription that keeps track of who said what
Upload an interview and get every word with a speaker label and a timestamp at the start of each paragraph, plus a CSV with one row per speaker turn for NVivo, ATLAS.ti, MAXQDA or Excel. A neutral summary, the topics in order and notable quotes are written from the transcript; the AI does no coding or interpretation. An interview of up to 15 minutes is free in full; a longer one costs $5, paid once, for up to 6 hours.
- Interviews up to 6 hours for one payment (12 hours on Unlimited), files up to 2 GB
- A speaker label on every paragraph, renamed to Interviewer, Participant 3 or real names with Find and replace
- A timestamp on every paragraph, counted from the start of the recording
- Speaker-turn CSV with File, Time, Speaker, Text and Translation columns, one row per paragraph
- Optional translation under each paragraph and in its own CSV column
- Neutral summary, topics in order and word-for-word quotes; no coding or themes by the AI
- Up to 15 minutes free in full, first part of longer interviews free · whole interview $5, one payment
What you get back
A transcript built for analysis and quoting: every turn labeled, every paragraph timed, and a short neutral overview on top. Edit it on screen and every download follows your edits.
- The whole interview word for word, in the language spoken, never summarized; unclear words are marked [inaudible].
- Every paragraph opens with its time and its speaker, for example [1:04:37] Speaker 2:, and a new paragraph starts whenever the speaker changes.
- If you choose a translation, it appears under each paragraph, with the original first.
- A neutral summary of the interview in one or two paragraphs, and each speaker label with an apparent role (interviewer, participant) only where the content makes it clear.
- The questions or topics in order, each with its time and a one-sentence neutral summary of the answer.
- Notable quotes word for word, with speaker and time, ready for a report, article or thesis.
- Word, PDF and TXT (overview first, then the full transcript), the speaker-turn CSV, and SRT and VTT captions.
An interview in a chatbot, and in this tool
A chat assistant that accepts audio can handle a short, clear exchange. A full interview has to keep every turn, every speaker and every time in order, which is where a single long reply falls apart.
| What you need | A chatbot | This tool |
|---|---|---|
| A short clip with one clear voice | Usually fine, if your plan accepts audio | Also fine, and free in full up to 15 minutes |
| A full interview of an hour or more | Audio length limits; long transcripts get shortened or cut off | Up to 6 hours per job, every minute written out |
| Who said what | Turns blur together when a long reply is condensed | A speaker label at the start of every paragraph |
| Going back to the audio | No reliable timestamps on a long file | A timestamp at the start of every paragraph |
| Quotes for a report or article | Can be paraphrased in a summary | Word for word, with speaker and time |
| Import into analysis software | Text in a chat window to copy out | Speaker-turn CSV plus Word, PDF and TXT |
How to transcribe an interview with speaker labels
-
1
Upload the interview
Drop the audio or video file (MP3, M4A, WAV, MP4 and more, up to 2 GB), or paste a public Google Drive, Dropbox or direct file link. If one session was saved as several files, add them together; they are joined in name order.
-
2
Check the first part and rename speakers
The opening minutes appear with a speaker on every paragraph. Use Find and replace to turn "Speaker 1" into Interviewer and "Speaker 2" into Participant 3 or a name.
-
3
Unlock and download
Pay $5 once for the whole interview, then download Word, PDF or TXT, or the speaker-turn CSV for your analysis software.
Why long interviews lose their structure in a chatbot
An hour of conversation runs to many thousands of words. Chat assistants that accept audio limit how long a file can be, free plans often cannot upload audio at all, and a transcript that long does not fit in one reply. What comes back is shortened, summarized or stops partway, and the first thing lost is the turn-by-turn structure: who asked, who answered, and when.
This tool cuts the recording into parts of up to ten minutes and transcribes each part in full, with an instruction to start every paragraph with its speaker. Timestamps are shifted by where each part starts and the parts are joined in order. The summary, topics and quotes are written afterwards, from the finished transcript only.
Who it is for
- Qualitative researchers and students transcribing interviews for a thesis or study
- Journalists who need exact quotes and the time each one was said
- Oral history projects recording long life-story interviews
- Asylum, immigration and legal interviews that need a working transcript and a translation
- User research and hiring interviews reviewed by a team
Speaker labels and the CSV: what to check
Speakers are labeled "Speaker 1", "Speaker 2" and so on; a real name is used only if the person is clearly named in the recording. The numbers are assigned per ten-minute part, so on a long interview the participant can become "Speaker 1" in a later part. Skim the labels before you download and fix them with Find and replace.
The CSV reads the label before the colon at the start of each paragraph into the Speaker column. Labels such as Interviewer, Participant 7, Respondent 2 or a name of up to three words are read correctly; a bare code like P07 is not, and stays in the Text column. Your on-screen edits and renamed speakers are included. If the job has several files, the File column shows which file each turn came from.
What the AI does not do, and accuracy
The AI does not code the interview, tag themes, interpret what anyone meant or judge a participant; that is the researcher's work. It does not time pauses or mark overlapping speech in the detail conversation analysis needs. The summary, topics and quotes are written by AI from the transcript, so check them against the audio before you cite them.
The transcript is made by an AI speech model and does well on clear speech. Background noise, people talking over each other and a microphone far from one speaker cause mistakes; check names and numbers. It is not a certified transcript or a sworn translation, so a court or immigration authority may require one from a certified transcriber or translator. Check that your consent forms or ethics approval allow transcription by an outside service.
Interview jobs people bring
Pages for particular kinds of interview. They use the same tool, with the same price and limits.
Asylum interview recording to text, with English
Upload the recording of an asylum or immigration interview in Pashto, Dari, Arabic, Tigrinya, Kurdish or anoth...
Read more →Court or tribunal hearing recording to text
Upload the audio of a hearing, a deposition or a recorded police interview and get a timestamped transcript wi...
Read more →Oral history interview transcription
Upload an oral history interview, a grandparent telling their life story or an archive tape, and get a verbati...
Read more →Interview transcripts ready for NVivo
Upload a research interview and get a transcript with a speaker label on every paragraph, as Word and TXT docu...
Read more →What it costs
Short interviews, first part
Free
An interview of up to 15 minutes is free in full, downloads and CSV included, no account. For a longer one, the opening minutes appear free on screen with speaker labels.
Whole interview
$5
One payment for one interview of up to 6 hours: labeled transcript, overview, Word, PDF, TXT, speaker-turn CSV, SRT and VTT. No subscription.
Unlimited
$19 / week
Renews weekly, cancel any time. Every tool on the site, recordings up to 12 hours per job, useful for a week of fieldwork.
Questions and answers
How do I transcribe an interview with speaker labels?
Upload the recording or a public Drive or Dropbox link and check the free first part, where every paragraph already starts with a speaker label. Rename the labels with Find and replace, then unlock the whole interview for $5 and download Word, PDF, TXT or the speaker-turn CSV.
Can ChatGPT transcribe a one-hour interview with speakers?
For a short excerpt it can. A one-hour interview often goes past the audio length limit, and a transcript that long does not fit in one reply, so it is shortened and the turns run together. This tool transcribes every ten-minute part in full with a speaker on every paragraph and joins them in order.
How do I get the transcript into NVivo, ATLAS.ti or MAXQDA?
Import the Word or TXT file as a document if you code line by line, or the CSV as a dataset if you want each speaker turn as a row with File, Time, Speaker, Text and Translation fields. The import steps differ between versions, so follow your program's help for importing documents or spreadsheet data.
Why did a speaker's number change partway through?
Speaker numbers are assigned within each ten-minute part, so the same voice can get a different number in a later part of a long interview. Rename the labels on screen with Find and replace before you download; the CSV and documents use your edits.
Does the AI code or analyze the interview?
No. It writes a neutral summary, lists the topics with a one-sentence neutral summary each, and picks quotes word for word. Coding, themes and interpretation are left to you.
Can it translate an interview?
Yes. Choose one of 77 translation languages and each paragraph is shown in the language spoken with the translation under it; the CSV gets the translation in its own column. The summary and topics are then written in the translation language.
Is it suitable for an asylum or court interview?
It gives a working transcript and translation to read and check, with a timestamp for every paragraph. It is not a certified transcript or a sworn translation, so an authority may ask for one made by a certified professional.
How much does it cost?
An interview of up to 15 minutes is free in full. A longer one is $5, paid once, for up to 6 hours. Unlimited is $19 a week, renews weekly, cancel any time, and covers recordings up to 12 hours per job.
Is my recording private?
Only you can open your job. The audio is sent to an AI model provider only to process it and is not used to train models. Jobs started without an account are deleted after 7 days, and you can delete a job and its files at any time.
Updated October 25, 2026 · APPFACT STUDIO LTD