Accountant: are you still replaying client voice notes just to catch the number?
The 18th, a little past seven in the evening. A voice note lands, 3 min 40 s long. The client talks about September revenue, mentions a new supplier, reads out an invoice number, remembers that his wife joined the company, and finishes by asking whether the filing can be brought forward. You listen to all of it. You write down 12,480. Then you are not sure whether it was 12,480 or 12,408, so you drag the bubble back to 2:10 and listen again.
Forty extra seconds. On one voice note. Now multiply that by a full client list in close week.
The size of the queue
In Brazil, where this extension was built, there are 549,002 active accounting professionals and 106,612 registered accounting firms (Federal Accounting Council, checked on 1 October 2026). Every one of those firms carries a client list. Not one of those lists agreed to send things by email.
On the other side of the chat, the volume. WhatsApp itself says its users send an average of 7 billion voice messages a day (WhatsApp, March 2022). Audio stopped being a quick aside. It became the channel where clients ask about a filing, confirm a deadline, describe the invoice that does not add up and mention they have moved office.
It makes sense on their end. The client is driving, standing behind a counter, hands busy. Talking beats typing. The cost never shows up for the person recording it. It shows up an hour later, for the person who received it, with another forty messages stacked in front.
Why a voice note costs more than it looks
There is a decent measure for this. A meta-analysis of 190 studies, covering 18,573 participants, measured how much an adult reads per minute: 238 words per minute for silent reading of non-fiction, 260 for fiction, and 183 words per minute when reading aloud (Brysbaert, Journal of Memory and Language, 2019).
Keep the last one in mind: 183. That is the order of magnitude of the spoken word. A voice note delivers content at the pace of whoever recorded it. Text delivers at the pace of whoever is reading, and a reader can skip the "hi, hope all is well over there", land on the line with the figure and move on.
Speed is the smaller part of the bill, though. What really weighs is everything that does not exist inside a sound file. You cannot search a word inside it. You cannot copy the tax ID and paste it into the system. You cannot attach the relevant passage to the client folder. You cannot ask a junior to double-check it without making them sit through the same 3 min 40 s, barking dog included.
What exists today, and where each route stops
No practice has been sitting still waiting for a fix. Everyone has worked out some arrangement. It is worth looking at where each arrangement hits the wall.
| Route | What it solves | Where it stops |
|---|---|---|
| Listen and take notes on the spot | Nothing to install, works on any device | Costs the full length of the audio and still asks for a second listen to confirm a figure |
| Play at 1.5x or 2x | Cuts a good share of the listening time | A figure said quickly gets worse at 2x, and none of it becomes text |
| The WhatsApp transcript itself, on the phone | Lets you read the message without pressing play, inside the app | One message at a time, on the handset, and the text never reaches the computer where you work |
| Download the file and send it to a transcription service | Returns the full text of the audio | Download, upload, wait, copy back, and client data travels through one more place |
| Transcribe inside WhatsApp Web, with an extension | Text appears under the audio, in the chat itself, while you work | Depends on the computer being on with WhatsApp Web open |
Notice the pattern in the first four. When text exists at all, it is not where the work happens. It sits on the phone, in your head, or in a notepad nobody opens again.
How it looks inside WhatsApp Web
ZapVox is a browser extension that works on top of WhatsApp Web, on the computer. Incoming audio turns into text in the chat itself, and that text then serves the rest of the day.
- Automatic transcription — switch it on per chat, for the clients who send audio every day. What arrives shows up as text under the bubble without you clicking anything.
- Glossary and names your way — teach it the words that only exist in your work: the filing codes, the payroll terms, the partner name that would come out misspelled. From then on they come out right.
- Bullet-point summary — the six-minute voice note becomes four lines: what the client asked for, the deadline mentioned, the figure quoted, the item still pending.
- Search inside voice messages — on Thursday someone asks who had mentioned the director salary. You type the word and ZapVox shows which chats it appeared in, with the passage around it.
- Export the chat transcriptions — that client conversation comes out as .txt and goes into their folder with everything else.
- Schedule messages and files — the chase for the missing document is set for the 15th at 9am, written today while the subject is fresh.
The shape of the week changes quietly. You open the chat with the text already there, glance at the summary, check the figure against the passage where it appears, and move on. The audio is still there, untouched, for when the doubt is serious. You can try this on your own client list before changing any process, because the trial opens everything for seven days.
Where this stops
The part no advert tells you.
- It lives on the computer. The extension runs on WhatsApp Web, in Chrome, Edge and Brave. With the browser closed nothing is transcribed, so anyone who handles everything from the phone will get little out of it.
- It does not touch your accounting system. ZapVox does not post entries, does not issue tax forms, does not sign anything and does not talk to your practice software. It hands you text, and the posting stays yours.
- Figures said fast can come out wrong. Audio recorded in the street, a quiet voice, a client talking over the radio: transcription gets the meaning right and can still miss a digit. Check the amount and the tax ID against the audio before using them. The tool speeds up reading, it does not take responsibility for the entry.
- A transcript is a working instrument. It exists to organise what was said and to find it later. It does not replace the invoice, the contract or the document, and it is not presented as proof of anything.
- There are quotas. The Free plan gives 10 transcriptions a day and up to 15 minutes of audio a day. Starter gives 60 a day and up to 30 minutes. Pro gives 150 a day and up to 150 minutes. Bullet-point summary, glossary and export sit in the paid plans and in the trial. On Free, search reaches the last 7 days.
One note from watching this land in a practice: the gain does not show up on day one. It shows up in week three, when someone asks what the client said about the August invoice and the answer comes back in fifteen seconds, passage on screen, instead of arriving after a "let me go and look for that" that disappears into the afternoon.
Seven days, no card
Every sign-up starts with 7 days of trial with every feature unlocked and no credit card. When the seven days end, the account drops back to the Free plan on its own and nothing is charged: there is no cancellation fee and no hidden subscription. The figures are on the plans page, and the full list of what the extension does is under features.
One close week is enough time to find out whether text under the audio changes anything in your day. If it does not, you spent seven days and no money. Install ZapVox for free.
In two lines
- The bottleneck was never the listening. It is turning what was said into something you can search, copy and file.
- The right place is where you already work. Text on the phone does not help someone closing the month at a desk.