Episode 42 went to tape on Tuesday night. The guest has gone home, the edit is done in the audio editor you always use, and it's due out Friday morning. The show, the episode and the guest are made up for this post. What's left is the writing and the packaging, which is where the hours go on a show made by one or two people.
Recording and publishing someone's voice is your job, and the agreement is best settled before they walk out the door. Ask the guest to agree in writing that you may record the conversation, publish it, and cut clips from it, and note anything they asked you to leave out. The Australian Law Reform Commission reported in 2014 that state and territory laws differ on whether a party to a private conversation may record it without the other people's consent, so check the rules where you record. General information, not legal advice.
Write the leave-out list somewhere you'll see it on Thursday. It matters again when the show notes get drafted, in case a section you cut survives in the transcript.
This workflow starts from a transcript you already have. Export one from the tool you edit or host in, as a text, Word or Markdown file, after the final cut so it matches what listeners hear. In the desktop Document editor, open Files & Export and choose Import file.... It takes .txt, .docx, .md and a few other text formats. A transcript holds your guest's words and name, and how Oppermind handles files you import is set out in our privacy policy.
Tidy it before anything else reads it. Fix the speaker labels so "Speaker 1" becomes a name, correct the spelling of the guest's name and any place or product they mention, and mark the sounds that carry meaning, like laughter or a clip you played.
That tidy transcript is worth publishing too. W3C's success criterion 1.2.1 (Level A) asks for an alternative to prerecorded audio-only content that presents equivalent information, and its Understanding page describes a transcript that identifies who is speaking and notes other significant sounds. Whether you're required to meet WCAG depends on who you are and who you publish for. General information, not legal advice. Even where nobody requires it, a transcript page gives search engines the words in your episode.
With the transcript open, ask Lato in chat for the notes. Lato reads and edits the document that's open beside the chat, so the request can be specific:
"From this transcript, write show notes for episode 42. A two-sentence summary that says what a listener will get from it, a short bio of Dana using only what she says about herself, five chapter headings in the order they come up, and a list of every book, tool or website mentioned. Leave out the part about her old employer."
Ask for the notes in a new document, or below the transcript, so the source stays beside them.
Then go through the draft against the transcript and your edit:
Export the finished notes as a Word file or PDF from the same Files & Export menu, or copy the text into your host's episode description box. Put the summary at the top and keep it short.
Most shows open with the same few lines every week, plus one that changes: "This is episode 42, with Dana Reyes on running a community radio station." In the Voice editor that intro is a project of separate lines, each with its own voice, speed, pause after it and an Instructions box for direction.
Keep one intro project as the master. Each week, edit only the episode line. Its status drops back to draft, and the other lines keep the audio they already have. Press that line's Generate button and play the project through with the pauses in place. If the line has to sit closer to the music, change its pause rather than the speed, because speed is passed to the voice as a direction about pace and the exact length is left to the voice.
Then use Export WAV from the More menu. The export stitches the lines with their pauses and levels into one file, which you drop into your audio editor ahead of the interview. WAV suits a file that will be mixed and re-encoded later.
Each generation counts toward the plan's daily voice limit, including a second take of the same line. On current plan terms, Free includes 5 voices a day, so a weekly show that re-voices one line spends one of them.
The Voice editor's voices are stock voices from a fixed list. The editor doesn't clone voices, so there's no way to upload a sample of yourself or a guest and have the editor copy it. If the intro should be in your own voice, use Record from microphone to add your line, and it plays in order with any generated ones. If a synthetic voice opens the show, say so in the show notes, so nobody mistakes it for a person on your team.
The intro usually sits over a few seconds of music. The AI music and MIDI maker post covers briefing the Music editor, fixing drums and notes, and which export to pick. For a podcast sting the useful sum is length. A project made from chat runs from one to eight bars, and in 4/4 at 120 BPM each bar lasts two seconds, so four bars gives an eight-second sting. At 90 BPM the same four bars run about 10.7 seconds. Export WAV and lay it under the intro in your audio editor. Don't ask for a copy of another show's theme or a song you know.
The directories show your art as a small square, often next to dozens of others, so the size and the text both have to survive being shrunk. The two big directories publish their own guidance.
| Spec | Apple Podcasts (Show Cover page, undated, checked 17 Nov 2026) | Spotify (show art guide, 2 Aug 2019, checked 17 Nov 2026) |
|---|---|---|
| Size | 3000 by 3000 is the standard. 1400 by 1400 to 3000 by 3000 is accepted through an RSS feed, largest preferred | Between 1400 by 1400 and 3000 by 3000 |
| Shape | Square | 1:1 |
| File type | PNG or JPG | Not stated on that page |
| Background | Solid, no transparency | Suggests solid colours or gradients, with a block of colour behind text |
| Text | Title clear and prominent, limit extra text and logos, keep key elements inside the safe area | Not stated in specs |
Their pages change, so check both before you submit.
In the desktop Image editor, open your show's cover image or ask Lato to generate a background. Add the episode number with the Text tool (it has font, size, fill and stroke settings) and keep the show title as the largest thing on the square. Use Crop and watch the live pixel readout until width and height match, then Resize to 3000 by 3000. Export from Files & Export as JPG, which carries no transparency, or as PNG from a fully opaque image.
Zoom out until the art is about the size of a thumbnail on a phone. If you can't read the episode number at that size, make it bigger or cut words.
If the art includes a generated picture of a person, it's an illustration. Don't make it look like the guest unless you're using a photo they've agreed to, and don't present a generated face as anyone real.
By Friday morning the episode folder holds the mixed audio from your editor, the show notes as text, the transcript, and one square image. Paste the notes into your host, upload the art, link the transcript page in the notes, and schedule it. Keep the notes and transcript after the episode goes out, because one day a listener will ask which book Dana mentioned.
Import the transcript into a document and ask Lato to draft the show notes from it.
The episode file your audio editor exports. The intro and sting WAVs from Oppermind go into that edit first, where you set their levels against the interview.
Fix names and speaker labels by hand first, then ask Lato to correct obvious mishearings using the episode topic as context. Mark anything you're unsure of and check it against the audio.
It's optional. Some shows keep one cover and change only the episode title. If you do make per-episode art, the same size and text checks apply.
Record her line from the microphone in the Voice editor, with her agreement, or record it in your audio tool and leave the Voice editor for the lines nobody reads.
Oppermind is not affiliated with, endorsed by, or sponsored by Apple, Spotify, W3C or the Australian Law Reform Commission.