Run the interview. Dalm keeps the quotes.

Qualitative research transcription is turning recorded interviews and focus groups into text you can read closely, quote and analyse. For a study, the transcript has to keep who said what, let you find a passage again across dozens of interviews, and fit what participants were told about their data. Dalm, a Mac app, records the interview or takes the file you have, recognises the speech on your Mac and sends no audio anywhere; the transcript is synced to our servers automatically, and a language model reads it to write the summary. There is a Free plan.

What a qualitative study needs from a transcript

A transcript for a study is data, not a record of a meeting. It has to show who is speaking on every line, so an interviewer’s question is never read as a participant’s answer. It has to be exact enough to quote. And it has to be findable again weeks later, when the analysis turns up a theme and you need every passage that touches it. Some methods keep every pause and false start, others clean the words up for reading — verbatim transcription sets out the difference. Software does the first pass either way; reading it against what you remember of the interview stays with you.

Recording research interviews: a call, a room, a file

On a call, Dalm takes your microphone and the call’s sound as two streams, so your questions come out as yours and the participant’s answers as another voice. In a room, one microphone hears everyone, and the voices are told apart on your Mac before the text is synced to our servers; focus group transcription works the same way, with more voices to name. An interview recorded on another device comes in as a file of up to four hours, and the original stays where it was. How to set each of these up, and what to check before you press record, is in the guide to transcribing an interview. If you are a product manager running your own interviews, the rest of your meetings are on the page for product managers.

Interviews, usability tests and focus groups: what you get after each

Dalm works out from the conversation what kind of session it was, and shows it at the top of the meeting. You cannot pick or change it, and when the app is not sure, it treats the recording as a general meeting. For the three kinds of session a study is usually made of, this is what comes after:

The session	What you get after it	What it is for in a study
A research interview	A one-screen summary of the session, and a write-up laid out for a research interview	The summary helps you decide which sessions to read closely first; the write-up is a starting point you check against the transcript
A usability test	A one-screen summary, a list of follow-ups — who and by when — and a write-up laid out for a usability test	The follow-ups are what the team has to fix or check after the session
A focus group	A one-screen summary of the session, and a write-up laid out for a focus group	A first pass over what the group said, to check against the transcript

Every one of them also gives you the transcript, line by line with speakers and times, and a summary in which each point keeps a quote from the transcript. The summary is written automatically; the other documents are one click away. If every session in your study needs the same write-up, describe it once in your own words as a template and tick the box to build it for each new meeting.

What happens to a participant’s words after the interview

The participant’s voice is recognised on your Mac and is never sent anywhere as audio. The audio is deleted once the text exists. The words go further. The transcript, line by line with its times and voice numbers, is synced to our servers automatically, and no setting turns the sync off. A language model, run by a third party, then reads that text automatically to write the summary, and again to answer the questions you ask about the interview. What goes where, piece by piece, is on privacy; what changes when a service takes the recording itself is on local and cloud transcription compared.

Participant codes instead of names

Naming the voices is optional, and the name you give is plain text that is synced with the transcript. So if a participant should appear as P07 rather than by name, give the voice that code. Until you answer, a voice shows the app’s guess or a number; in the naming step a guess may be marked “Suggested by Dalm — check before you agree.” In the transcript and the summary a guess looks like any other name until you confirm or change it. Editing a line changes what you see, with the original one click away, but the copy on our servers stays as it was recognised: that copy is what the language model reads. So if a participant says their own name or someone else’s, that sentence is there as it was spoken. A voice is not carried over between recordings either: in a follow-up interview, give the code again.

Finding and checking quotes across a project

Tags keep the interviews of one project, site or participant group together, and the list of meetings filters by tag, person and period. The search field looks through the transcripts, names and documents of every meeting at once — your own notes are not searched — and ⌘F finds a word inside one transcript. Any line can be copied with its speaker and time (“Copy with name and time”). In the meeting summary, each point keeps a quote from the transcript. Every quote shows you where it was said — open it and the transcript opens at that place.

Everything you say lands in one archive you can search: meetings, phone calls, conversations in the room, and what you dictate to a model. Questions to the assistant are asked of one interview at a time; asking one question across the whole project is in development.

What to tell participants about the recording

Say at the start that you are recording. Dalm does not join the call and does not tell anyone that it is recording; the app reminds you of this before your first meeting. If your information sheet describes what happens to recordings and transcripts, this is what happens with this app: the audio does not leave your Mac, the transcript is kept on our servers until it is deleted, and a language model run by a third party reads it to write the summary. Whether that fits what your participants agreed to, and what your study’s rules require, is for you and whoever oversees the study to decide. The rules on recording consent differ by place.

When Dalm is not the tool for a study

• When transcripts may not leave your computer or your institution’s systems, even as text. The transcript is synced to our servers and read by a language model, and no setting keeps it on the Mac.
• When the audio has to be kept to check the transcript by ear. The audio is deleted once the text exists. A file you bring in stays where it was, so keep your own recording.
• When the analysis needs coding. Across interviews the app gives you search and tags; coding and theme analysis are the work of analysis software — programs are compared on the page about transcription software.
• When a conversation is very short: with fewer than 100 words, a recording is not counted as a meeting.
• When interviews are recorded on a phone or on Windows. Dalm is a Mac app; a recording made elsewhere can be brought in as a file.

Free for one researcher, Team for a research group

Recording interviews, the transcripts and the summaries come with the Free plan, and Free is a plan rather than a trial. Dalm needs a Mac with Apple Silicon · macOS 15.0 or later and an account; while it is not public, the way in is the waitlist. What the paid plans add is on the pricing page. A research group can work in one team: the dictionary keeps participants’ jargon and site names spelled the same way in everyone’s transcripts, and tags and templates are shared. Who in a team sees which meetings is set out on working as a team — read it before putting a study there.