How to Fix Dictation That Mangles Names and Brands
Dictation is fast until it hits a name it does not recognize. Then a client called Siobhan becomes "Shivon", a product called Zscaler becomes "the scaler", and you spend more time fixing text than you saved. Here is how to make voice to text on Mac spell your names and brands right.
Key takeaways
- Names and brands break dictation because speech models guess the nearest common word.
- A custom dictionary is the real fix: add the exact spelling once and it sticks.
- On-device AI cleanup applies your dictionary and prompt so text comes out polished.
- BlaBlaType does this locally, so corrections work offline and your audio stays on the Mac.
Why dictation gets names and brands wrong
Speech-to-text models are trained on huge amounts of everyday language. That makes them great at common words and predictable phrasing, and weak at anything rare: personal names, company and product names, acronyms, medical and legal terms, and internal jargon. When the model hears a sound it does not have a confident word for, it does not leave a blank. It substitutes the closest common word it does know.
That is why "Xero" turns into "zero", "Kubernetes" turns into "cooper netties", and a colleague named Aoife turns into "ee fa". The audio was fine. The model simply never learned that spelling. This is the same root cause behind a lot of dictation frustration, and it is separate from the reliability issues covered in our guide to Mac dictation not working.
The fix: a custom dictionary plus AI cleanup
There are two layers that solve this together. The first is a custom dictionary: a short list of the exact spellings you care about, so the app knows that this particular sound means "Zscaler" and not "the scaler". The second is on-device AI cleanup, which takes the raw transcript and tidies punctuation, filler and grammar while respecting the names in your dictionary and any custom prompt you set.
Neither layer alone is enough. AI cleanup cannot invent the correct spelling of a name it has never seen, and a dictionary alone still leaves you with messy, unpunctuated speech. Used together they are reliable. Follow the steps below to set it up.
List the names that keep breaking
Spend two minutes writing down the people, companies, products and acronyms you dictate often. These are the words your current tool gets wrong the most.
Add them to the custom dictionary
Open the dictionary in BlaBlaType and enter each one with its exact spelling and capitalization, for example "Aoife", "Zscaler" or "SaaS".
Turn on on-device AI cleanup
Enable AI cleanup so the app fixes punctuation and filler while applying your dictionary. Add a custom prompt if you want a specific tone or format.
Dictate a test sentence and confirm
Speak a sentence packed with your tricky names. If one is still off, adjust the dictionary entry once. After that it stays correct.
Because BlaBlaType runs speech recognition and the dictionary entirely on your Mac, these corrections keep working offline and your audio never leaves the device. That is a real difference from cloud tools, and it matters if you dictate names under an NDA. If your underlying issue is that dictation cuts out mid-sentence rather than misspelling, see why Mac dictation stops after a few seconds.
Apple Dictation vs a custom dictionary tool
Apple's built-in Dictation is convenient and free, and you can add words to the systemwide text replacements, but it does not offer a dedicated dictation dictionary or AI cleanup that reliably enforces spellings. Here is how the approaches compare for names and brands.
| Capability | Apple Dictation | BlaBlaType |
|---|---|---|
| Custom dictionary for names | Limited | Yes |
| On-device AI cleanup | No | Yes |
| Custom AI prompt | No | Yes |
| Works fully offline | Mixed | Yes |
| Works in any app | Yes | Yes |
Apple's own Dictation guide and its on-device dictation notes are worth reading if you want to squeeze the most out of the built-in option first. When that still is not enough for your names, a dedicated dictionary is the upgrade.
Make dictation spell every name right
Add your names and brands once, get AI-cleaned text, and keep every word on-device. No card needed for the trial.
Download for macOSA checklist to keep names accurate over time
Names and brands change: new clients, new products, new acronyms. Run through this short checklist when you notice a word slipping, and dictation stays reliable.
Accurate-names checklist
- Add each recurring name, brand and acronym with its exact spelling.
- Match capitalization exactly, for example "iOS", "SaaS" or "GitHub".
- Keep on-device AI cleanup enabled so punctuation and tone stay clean.
- Add a custom prompt if a project needs a specific style or format.
- Test a sentence full of tricky names after any big change.
- Review the dictionary monthly and remove entries you no longer use.
With those habits, most people keep dictating at the pace that makes voice worth it in the first place. Remember that most people speak around three to four times faster than they type, so once the names come out right, dictation is simply faster writing. If dictation still drops out on you, our guide on Mac dictation that keeps stopping covers the reliability side.
Frequently asked questions
Why does Mac dictation misspell names and brands?
Speech models are trained on common words, so unusual names, product names and jargon fall outside their vocabulary. The model guesses the nearest common word, which is why a name like Siobhan or a brand like Zscaler comes out wrong.
How do I add custom words to dictation on Mac?
Use a tool with a custom dictionary. In BlaBlaType you open the dictionary, add the exact spelling of each name, brand or acronym, and the app matches your speech to those entries so they come out correct every time.
Can AI cleanup fix names in dictation automatically?
Partly. On-device AI cleanup fixes punctuation, filler and grammar, and it can respect a custom dictionary and prompt so known names and brands are spelled correctly. It cannot invent the right spelling of a name it has never seen, which is why a dictionary matters.
Does fixing dictation names work offline?
Yes. BlaBlaType runs speech recognition and its custom dictionary entirely on-device, so name and brand corrections work without an internet connection and your audio never leaves the Mac.
Will a custom dictionary slow down dictation?
No. A custom dictionary is a small list the app checks as it transcribes. It adds no noticeable delay and dictation stays as fast as normal, which for most people is three to four times faster than typing.