← Back to blog

Content Marketing · November 16, 2026 · 8 min read

How to Turn a Voice Memo Into a Blog Post

The transcript is never the problem. Ten minutes of talking holds about three posts of material; the work is splitting and cutting it, not writing it.

By FluxWriter

How to Turn a Voice Memo Into a Blog Post

Turn a voice memo into a blog post and the hard part is never the transcript. It is deciding which of the three posts inside it you are writing. Ten minutes of talking produces roughly 1,400 words, and those words usually answer three different questions with no headings, no order and a lot of restarts. Speech-to-text is free and usually accurate enough; what is left is structural work, and structural work is where most attempts stall. This covers how to record so the transcript is usable, how to split one memo into separate posts, and the single instruction that stops a rewrite from flattening your speaking voice into generic prose.


A Ten-Minute Memo Usually Contains Three Posts

Assume a normal speaking pace of about 140 words a minute. Ten minutes gives you around 1,400 words of transcript. That text almost never holds together as one argument, because you were not building one. You were answering whichever question came to mind next, and then the one it reminded you of.

That is the good news. Three loose answers are worth more than one forced essay, and each is the spine of a separate post, already carrying the example and the number you would never have bothered to type. Pick one, park the other two in a file, and one recording session has covered the next two months of publishing.

The failure is trying to keep all of it. A memo that wandered from pricing to onboarding to a complaint you got last Tuesday becomes a post that wanders the same way, and the reader who came for the pricing answer leaves in the onboarding paragraph. Splitting is not tidying. It is the difference between three posts that each answer something and one that answers nothing.

Record for the Transcript, Not for Yourself

Say the question out loud before you answer it. "Why do most quotes for this work come back too high" is a sentence you can lift straight into the draft as a heading, and it forces you to answer one thing at a time. Five questions, asked aloud, gives you a transcript that is already sorted into five blocks.

Talk to one person. Not an audience, not "customers" as a category, but someone you dealt with last month. Second person arrives on its own that way, and so does the level of explanation, because you already know what that person does and does not understand. Talking to nobody in particular produces the flat institutional register that makes the rewrite harder later.

Say the numbers out loud. A specific you do not say does not exist, and you will not add it back while editing, because by then you are tidying rather than remembering. Prices, timeframes, what the part actually costs. If you catch yourself saying "quite a lot", stop and give the figure.

When you lose the thread, say "scratch that" and start the sentence again rather than trailing off. A transcript keeps both attempts, and deleting a false start takes two seconds, while reconstructing an abandoned half-thought takes minutes of listening back. Do not read from notes either. Keep five prompts on paper, glance at them, and improvise, because the moment you read aloud you lose the thing that made speech worth using. As a rough rule, a quiet room and a phone about a foot away is the difference between a transcript you can edit and one you have to re-listen to.

The First Pass Is Cutting, Not Writing

Open the transcript and delete before you write a word. The first sixty to ninety seconds usually go, because they are you warming up. Then the repeats: speech circles back and makes the same point three times, since you cannot see that you already made it. Then the filler, the "so basically" and "at the end of the day" that carry nothing on the page.

Assume roughly half of it survives. That leaves about 700 words of substance from the 1,400 you recorded, and that 700 is split across the three topics you drifted between, so the post you actually chose is carrying something like 230 words.

Two hundred and thirty words is not a blog post, and this is where people decide the method failed. It is the part you could not have produced at a keyboard: the argument, the example, the number. The other 800 or so words are framing, transitions, and the context a reader needs before your example makes sense. That part is ordinary writing, and any writer or tool can do it without you.

The Time Budget Is About Ninety Minutes, Not Ten

Here is how the time tends to split for one post built from one memo. The figures are illustrative, based on a ten-minute recording at roughly 140 words a minute.

Stage Rough time What you end up with
Recording 10 min About 1,400 spoken words, three topics tangled together
Transcribing 2 min The same 1,400 words as raw, unstructured text
Splitting by topic 10 min One post chosen, two parked for later
Cutting 15 min About 700 words of substance, roughly 230 on your chosen topic
Drafting and expanding 30-45 min A draft of 1,000 to 1,300 words
Reading it aloud 10 min A publishable draft

That totals somewhere between 75 and 95 minutes, of which the recording is ten. The method does not beat typing on speed. It wins because the hardest ten minutes are ones you can do in the car park after a customer call, when the specifics are fresh and a blank document would have got nothing out of you.

It also scales badly. Thirty minutes of recording rarely gives you three times the material, because past the fifteen-minute mark most people are restating what they already said. Two ten-minute memos on two days beat one long session.

One Instruction Keeps Your Voice From Being Flattened

Most rewrites fail the same way. You paste the transcript into an AI tool, ask it to turn this into a professional blog post, and get back something fluent, evenly paragraphed and anonymous. Nothing in it is wrong. Nothing in it sounds like you either, and the specifics have been softened into sentences any competitor could have published.

The instruction that avoids this is a restriction rather than a request. Give it roughly as: use my words, you may delete, reorder and join sentences, do not replace my phrasing with your own, and where a sentence is unclear leave it alone and flag it in brackets for me.

The flag is the part that matters. Unclear sentences in a transcript are almost always where you said something specific and only half-formed it, which is exactly the material worth keeping. A default rewrite smooths those into a generic line and you never notice what was lost. A flag hands them back to you to finish in your own words, which takes a minute each.

Then check the output for the tells. Every paragraph the same length is one. Words you have never said out loud in your life are another, along with three adjectives where you used one. Speech is uneven, a four-word sentence followed by a thirty-word one, and that unevenness is what readers recognise as a person. If the draft reads smoother than you talk, pulling three or four of your original sentences back in is usually enough to fix it.

Speech Gets Specifics That Typing Talks You Out Of

Ask an owner in writing why an installation takes six weeks and you get a paragraph about a thorough process and attention to detail. Ask the same owner out loud and you get three weeks waiting on the electrician, one week for a part that ships from abroad, and a fortnight in the middle where nobody can start because the customer has not chosen a colour.

The second version is the post, and it rarely survives typing. Typing runs slower than thinking, which gives the self-conscious part of your brain time to generalise everything into safety. Speech runs ahead of that editor. This is the real reason to record a memo even when you could have written the post yourself in an afternoon.

A memo will also produce things you should not publish, which is worth a deliberate check rather than a skim. Named clients, real figures, a competitor discussed frankly, a complaint detailed enough to identify who made it. Before any of that goes live, the question to ask is what your agreement with that client says about confidentiality, and whether they would recognise themselves in the paragraph. Check, rather than assuming the story is yours because it happened to you.

FAQ

Do I need a paid transcription tool?

Usually not. The voice recorder and dictation already on your phone are generally adequate for one person in a quiet room, and accuracy matters less than people expect, because you are deleting about half the words anyway. Where free tools tend to struggle is proper nouns and trade jargon, so keep a short list of the terms you always say and run a find-and-replace before editing.

How long should the memo be?

Eight to twelve minutes is a reasonable starting point. Under five minutes there is rarely enough raw material to survive the cutting stage, and past fifteen most people repeat themselves rather than add anything. If you have more to say, record a second memo on a different day.

Will a post that started as a voice memo read as lower quality?

Not because it was spoken. The published page is text, and a reader is judging whether it answers the question they arrived with. The risk is publishing something too close to the transcript, where the repeats and the warm-up survive and the piece reads as unedited rather than as spoken.

The Practical Takeaway

Write down the five questions customers ask you most often, in their words rather than yours. Take the first one and answer only that question for ten minutes, saying it aloud before you start and giving a real number every time you are tempted to say "quite a lot". Transcribe it, delete the first ninety seconds and every repeat, and see how much substance is left before deciding whether the method suits you. Pick the question this afternoon and record the answer on your phone before you talk yourself into planning it instead.

Once the two parked posts are written up too, FluxWriter's manual uploads (Starter and up) can take all three .docx files at once and hold them as drafts, so each one is a single Publish click in the week it is due.



← All posts