Transcribe an interview for research, verbatim first

To transcribe an interview for research, record it or upload the audio file, then choose Quick Note — sAInote's verbatim mode, which returns the transcript word for word with nothing added and nothing removed. sAInote is an AI voice-to-text app for the web, iPhone, iPad and Mac. Recordings up to 20 minutes are processed immediately on every plan; interviews between 20 minutes and 2 hours are processed in the background and require Pro. Two hours is the hard cap per recording, and uploads must be 64 MB or smaller.

Transcribe an interview 15 minutes free, one time. No password — a 6-digit emailed code, or Apple or Google.

What you get from an interview recording

Quick Note: word for word: Quick Note returns the transcript exactly as spoken — nothing added, nothing removed, no tidying. It is the mode to use when you need quotable text you can attribute, and it is available on every plan, guests included.
Smart Summary as a second pass: Smart Summary is a separate, AI-edited pass: filler removed, half-finished sentences repaired, headings and lists added, facts unchanged. Run it alongside Quick Note when you want a readable digest — never instead of the verbatim record.
Speaker separation on long recordings: On long recordings, speakers are separated so interviewer and interviewee stay distinguishable in the transcript. Summarisation is hierarchical too, so detail from the final twenty minutes survives as well as the opening.
Special Names for spelling: Special Names is a personal spelling glossary on Pro. Add your interviewee's name, their institution and any technical vocabulary, and those words are spelled correctly rather than guessed by ear. It never adds or rewrites words.
Searchable history across devices: Signed-in accounts keep a searchable note history across web, iPhone, iPad and Mac. Search matches both the transcript and the processed output, so you can find a phrase months later. Guests get no history.
Edit & Translate for cross-language work: Edit & Translate works in Smart Summary mode only. It edits and translates in one pass into roughly 35 target languages, translating the cleaned output rather than the raw transcript, and remembers the language you chose.

How to transcribe a research interview

1

Record or upload the interview

Record straight into the browser, the iPhone or iPad app, or the Mac app — Quick Capture is Control+Option+Command+N. Or upload an existing file of up to 64 MB.
2

Add your names to the glossary first

On Pro, enter the interviewee's name, their organisation and recurring technical terms in Special Names, so those words are spelled the way you wrote them rather than guessed by ear.
3

Choose Quick Note

Select Quick Note before processing so the output is verbatim. Smart Summary is the other option, for when you want a readable digest rather than a verbatim record.
4

Let long interviews process in the background

Anything under 20 minutes comes back immediately. Between 20 minutes and 2 hours, Pro processes in the background — start it, close the tab, come back to the transcript.
5

Check the transcript against the audio

Play the recording back against the text before you quote it. Correct anything the audio disagrees with, then code or cite from the version you verified.

Verbatim first: how to transcribe an interview for research

An interview transcript is evidence. A journalist has to attribute a sentence to the person who actually said it, and a graduate student coding thesis interviews has to be able to show that the text matches the recording. That is why sAInote, an AI voice-to-text app, has a verbatim mode and asks you to choose it deliberately.

Quick Note is that mode. It returns what was said, word for word, with nothing added and nothing removed — no smoothing, no reordering, no filler cleanup. It is available on every plan, including guests recording in the browser without an account.

Smart Summary is the other mode, and it is genuinely a different thing: an AI-edited version that strips filler, repairs broken sentences and adds paragraphs, headings and lists, while leaving the facts alone. It is useful for a research memo or a fast read-through for an editor. It is not the record. Choose it when a readable digest is what you need, and quote only from a Quick Note transcript.

Usage is measured in minutes of audio: a 3-minute recording costs 3 minutes, whichever mode you choose.

Long interviews: the 20-minute line and the 2-hour cap

Research interviews routinely run past 20 minutes, so here is the boundary stated plainly. Recordings up to 20 minutes are processed immediately on every plan, guests and free accounts included. Recordings between 20 minutes and 2 hours are processed in the background and require Pro. Two hours is the hard cap for one recording — anything longer is rejected, so a three-hour focus group has to be split into separate files before you upload it.

Background processing means you start the job and leave. You do not watch a progress bar for a 90-minute interview; the transcript appears in your history when it is ready, on whichever device you signed in from.

Long recordings also get two things short ones do not need. Speakers are separated, so the interviewer's questions and the interviewee's answers stay distinguishable instead of running together in one block. And summarisation is hierarchical: the audio is processed in chunks and then combined, so what someone said in the last fifteen minutes carries as much weight as their opening answer — often the point in an interview where people finally say the thing you came for.

Names, terminology and languages

Names are where automatic transcription usually fails an interview. A surname, an institution, a drug name or a piece of field-specific jargon gets rendered by sound, and you lose an evening fixing the same word forty times. Special Names, available on Pro, is a personal spelling glossary: you add the words that matter — your interviewee's name, their organisation, product names, technical vocabulary — and they are spelled the way you wrote them rather than guessed at. It is a spelling reference only. It never rewrites what was said and never adds words that were not spoken, and your glossary stays private to your account.

Language coverage is wider than the interface. The interface exists in 11 languages — Arabic, Bengali, English, French, Hindi, Portuguese, Russian, Simplified Chinese, Spanish, Turkish and Urdu, with Arabic and Urdu rendering right-to-left — while speech recognition covers considerably more.

If you interview in one language and write in another, Edit & Translate handles roughly 35 target languages in a single pass. It works in Smart Summary mode only, and it translates the edited output rather than the raw transcript. For attributable quotes, keep the original-language Quick Note transcript as your source.

What you still have to do yourself

sAInote produces a transcript. It does not certify one. There is no notarisation, no certified-transcript stamp and no claim that the output is admissible as evidence or accepted anywhere as a verified record. If your ethics board, court or publisher requires certified transcription, this is not that.

Three jobs stay with you. First, verification: play the audio back against the text before you quote or code anything, especially where the recording is noisy, people talk over each other, or a speaker uses an unfamiliar accent or dialect. Automatic transcription gets things wrong, and a wrong word inside quotation marks is your problem.

Second, consent and anonymisation. Obtaining informed consent to record, and removing or pseudonymising names, places and identifying details before you share or archive a transcript, is the researcher's responsibility. Software cannot tell which small detail identifies someone in a small field.

Third, analysis. A transcript is raw data. Coding it, building themes, checking agreement between coders and writing up findings is the actual research. sAInote replaces the typing, not the thinking, and it is not a substitute for qualitative analysis software.

Interview transcription questions

Can sAInote give me a word-for-word interview transcript?

Yes. Quick Note is a verbatim mode: it returns the transcript exactly as spoken, with nothing added and nothing removed. It works on every plan, including guests without an account. Choose it explicitly — the AI-edited Smart Summary mode is a separate option, useful as a digest but not as the record you quote from.

How long can an interview recording be?

Two hours is the hard cap for a single recording, and anything longer is rejected. Up to 20 minutes is processed immediately on every plan. From 20 minutes to 2 hours, processing happens in the background and requires Pro. Split longer sessions, such as focus groups, into separate files before uploading.

How much can I transcribe for free?

15 minutes of audio, one time — not 15 minutes per month. That applies both to guests recording in the browser and to free accounts, and both get Quick Note and Smart Summary. Pro includes 360 minutes, six hours, every month, plus background processing for interviews over 20 minutes and Special Names.

Does it separate the interviewer from the interviewee?

Yes. On long recordings, speakers are separated so questions and answers appear as distinct turns rather than one undifferentiated block. That matters most for interviews running past 20 minutes, where a single wall of text is hard to code, quote, or read back against the audio.

Can I transcribe an interview conducted in Turkish or another non-English language?

Yes. Speech recognition covers considerably more languages than the 11 the interface is offered in, and the interface itself includes Turkish, Arabic, Bengali, English, French, Hindi, Portuguese, Russian, Simplified Chinese, Spanish and Urdu. Record in the language of the interview and keep the Quick Note transcript in that same language for quoting.

Can I translate an interview transcript into another language?

Yes, with Edit & Translate, which works in Smart Summary mode only. It edits and translates in one pass into roughly 35 target languages and remembers the language you picked. Note what it translates: the cleaned, edited output, not the raw verbatim transcript. For attributable quotes, work from the original-language Quick Note.

Will my interview recordings stay private?

Yes. Audio is held in private cloud storage — never public, never indexed, never shareable by URL — and cleaned up on a schedule after processing. You can delete your account in the app, which anonymises the data. Requests made through the REST API store neither audio nor text.

Is the transcript certified or court-admissible?

No. sAInote does not provide certified, notarised or legally admissible transcription, and makes no claim of academic or institutional certification. If your ethics committee, court or publisher requires a certified transcript, use a service that offers one. Verify any transcript against the original audio before you rely on it.

What audio files can I upload, and how large?

WAV, MP3, M4A, MP4, WebM, OGG and Opus, up to 64 MB per file. If a two-hour interview exceeds 64 MB, export it at a lower bitrate or split it. You can also record directly in the browser, in the iPhone and iPad app, or on Mac with the Control+Option+Command+N hotkey.

Keep the exact words

Start with Quick Note, add Smart Summary only if you want a readable digest beside it, and check both against the audio before you quote. The transcript is yours to verify, code and cite.

Transcribe an interview