For podcasters: a transcript of every interview, and show notes in your own format

Podcast transcription software turns an interview into a transcript, every line with its speaker and time, and then into what goes out with the episode: show notes, a description, quotes to promote it. Dalm is a Mac app that works alongside your recorder, not instead of it: it captures a remote interview without joining the call, or takes the finished episode as a file. Speech is recognised on your Mac and no audio is sent anywhere; the transcript, as text, goes to our servers automatically, and a third-party language model reads it to write a summary. Show notes come from a template you describe once. There is a Free plan.

What an interview show needs once the recording stops

An interview episode leaves work behind it that has nothing to do with the edit. The episode page needs a description that makes someone press play, show notes with the topics in the order they came up, and the books, people and links your guest mentioned; some shows put the full transcript beside the audio. The announcement needs two or three lines the guest actually said, word for word. And before any of it goes out, the guest’s name, and every name they dropped, has to be spelt right. A transcript is the raw material for all of that, not the show notes themselves. This page is about getting from one to the other without typing the interview out again. If you interview for an article rather than an episode, the page for journalists is about that work.

Record the interview as it happens, or bring in the finished episode

There are two ways in, and they give you different times. For a remote interview, run Dalm alongside the app you record in: it takes what your microphone hears and what the computer plays, nothing joins the call, and your guest sees no bot. Start it with ⌃⌥M or the Meeting button — it also offers to record on its own when Zoom, Microsoft Teams, Slack or FaceTime takes the microphone — and stop it yourself when you finish, because it does not stop when the call ends. In the same room, the voices are told apart on your Mac, and the text made from them is synced to our servers. Times in that transcript count from the moment Dalm started, not from the start of your edit. For times that match what listeners hear, bring in the finished episode instead: any audio or video file your Mac can play, up to four hours, and its times count from the start of the file. If host and guest were recorded on separate tracks, bring in the mixed episode as one file; getting a transcript of an episode file is set out step by step on its own page.

After an interview episode: what you get, and where show notes come in

Dalm reads each conversation, decides what kind of meeting it was and writes that at the top; you cannot pick or change it, and when it is not sure, it files the conversation as a general meeting. An interview recorded for a podcast or the press is one of the kinds it can recognise. When it does, you can open two documents with a click: a summary of the interview that fits on one screen, and the transcript laid out as a document, every line with its speaker and time. Whatever the label, the transcript is there and a summary is written without being asked; each point keeps the words it came from, and a name or a date nobody said is left blank rather than guessed. Fix any line before you use it.

Show notes are not one of those documents, so you set them up once. Describe yours in your own words as a template and tick the box to build it for each new meeting. For example:

Part of the episode pageWhat you might ask your template for
The episode descriptionTwo or three sentences on what the guest came to talk about, in the show’s voice
The show notesThe topics in the order they came up; the people, books and links the guest mentioned
The announcementThree short lines the guest said, word for word, each with who said it

The box applies to every meeting you record or bring in from then on, not only episodes, so if you also record other calls in Dalm, they get show notes too. For a single episode you can instead write what you want and get one document for that meeting. Read whatever it writes against the transcript before it goes up. What other kinds of interview leave you with is compared on the interview transcription page.

What happens to a guest’s words before the episode is out

Until the episode is published, what your guest said is unpublished, and a transcript of the raw recording also holds everything you will cut. Here is where it goes. Speech is recognised on your Mac, and no audio is sent anywhere. The audio is deleted once the text exists. An episode file you bring in is read where it sits, and no copy is made. The text is what leaves. The transcript is synced to our servers automatically — no setting keeps it on the Mac — and a language model from a third-party provider reads it to write the summary, your show notes and any document you ask for. There is no setting that keeps one episode away from the model. On our servers the text has no expiry date: it stays for as long as your account, or the team workspace it was filed in, exists, unless you delete it. What is kept where is set out on the privacy page.

Times that match the episode, and quotes you can trace

Every line of the transcript carries its time. In an episode you brought in as a file, the times count from the start of that file, so if it is the version you publish, a time in the transcript is the same point in that file; there is no setting to shift them. How times work in a transcript has the detail. In the summary, each point keeps the words it came from. Every quote shows you where it was said — open it and the transcript opens at that place. That is how you check a line before it goes into the announcement. Names need the same care: until you name a voice, the transcript and the saved files show the app’s guess, and a voice with neither a name nor a guess becomes “Speaker 1”, “Speaker 2”. Check the guest’s name before anything is published.

What to tell a guest about the recording

Dalm says nothing to the people on a call and adds nothing to it, so telling your guest is up to you; before your first meeting, the app reminds you. A guest who agreed to be on the show expects the episode to be recorded; a second program turning the conversation into text, and where that text goes, is worth a sentence of its own. For example:

“Alongside the recording for the show, I run an app on my Mac that turns our conversation into text. It keeps none of the sound, apart from a few seconds of each voice until I label it. The text is stored on the app’s servers, where an AI model run by another company reads it to write up notes.”

If a guest would rather it didn’t, record the episode as usual and bring in the edited file afterwards; then only what you publish becomes text in Dalm. The rules on recording consent differ by place, and the rules state by state are set out separately.

When Dalm does not fit how you make a show

  • When you are looking for the program that records or edits the episode, or cuts clips from it. Dalm keeps no audio, has no player and exports no sound; the episode stays in your recorder and your editor.
  • When you want to paste a link or an RSS feed. Nothing on this site takes one, and the app works from an episode file on your Mac.
  • When you expect show notes, a description and chapters to appear without describing them first. There is no show-notes document of its own: you write the template once, and you read what it produces. How it compares with a service that turns episodes into clips and posts.
  • When the transcript has to go up without a read-through. Names are guesses until you check them, and no accuracy figure is published.
  • When a guest’s words may not be held by an outside service before release, even as text. The transcript goes to our servers, and a language model reads it.
  • When you want the app to publish an episode page by link. Documents leave as text you copy, as Word, PDF or Markdown files, or through Share as text; putting them on your site is your step.
  • When a recording runs so long that its transcript is too large for our servers to accept: it then stays on your Mac only and gets no summary.
  • When you work on Windows or on an Intel Mac. Dalm needs a Mac with Apple Silicon · macOS 15.0 or later.

Free for one host, Team for a small studio

A show with a co-host, a producer or an editor can work in one Team workspace, with one dictionary for guests’ names and the show’s own terms, and one set of templates, so every episode’s show notes follow the same format. Who sees whose episodes is on the team page; what the paid plans add is on the pricing page. Dalm needs a Mac with Apple Silicon · macOS 15.0 or later and an account, and until it is public, the waitlist is the way in.

Questions

Does Dalm write show notes for an episode?
Once you have described them, yes. There is no show-notes document built in: after an interview you get the transcript and a summary. Write what your show notes contain as a template, in your own words, tick the box to build it for each new meeting, and from then on they are written after every meeting. Read them against the transcript before they go up.
Can I get chapter times for an episode?
The times are in the transcript: every line has one, and for an episode you brought in as a file they count from the start of that file, so if that file is the version you publish, they line up with it. Pick where each chapter starts from there.
Will my guest's name be right in the transcript?
Only once you have checked it. Until you name a voice, the transcript and the saved files show the app's guess, even if you skipped naming, and a voice with neither is saved as "Speaker 1", "Speaker 2". Rename the voices to fix a name; a line you edit is saved as edited.
Who can read an episode transcript before it is released?
A meeting in your own workspace is yours. In a team workspace, colleagues may open it depending on their role and the kind of meeting — see the team page. Either way, the transcript sits on our servers with no expiry date, and a third-party language model reads it to write the summary. The audio is not sent: speech is recognised on your Mac.
Can I publish the transcript or the show notes straight from the app?
No. Documents leave as Word, PDF or Markdown files, or through Share as text to Mail, Messages or AirDrop. Putting them on the episode page is your step.
My show is not in English. Will it work?
The default speech model covers English, Russian, German, French, Spanish and twenty more, a second one covers over a hundred languages, and you do not set the language. Summaries and documents are built for English and Russian: for a show in another language, the headings and labels the app fills in come out in Russian, while the points themselves should follow the language of the show, though no live recording has tested that. How a recording in another language comes out is covered on transcription in other languages.
Is there a free plan for podcasters?
Yes. It is a plan rather than a trial, and recording interviews, bringing in episodes, transcripts and summaries come with it; a studio adds shared terms and templates in a team. The paid plans are described on the pricing page. You need a Mac with Apple Silicon · macOS 15.0 or later and an account; until Dalm is public, join the waitlist.