Bold, italics, capitals: what text-to-speech doesn't hear
Bold and italics can be seen, not heard. What a synthetic voice really does with a visually stressed word, and how to write so your emphasis survives the move to audio.
When you set a word in bold, you're giving the reader's eye an instruction: slow down here, this word matters. The trouble is that the instruction is purely visual. A synthetic voice reads text, not formatting, and in the vast majority of cases your bold vanishes without leaving a trace on the ear. Understanding why saves you a disappointment, and more importantly it changes how you write anything meant to be listened to.
What happens between your editor and the voice
An article published on the web is HTML. Bold is one tag, italics another, colours and sizes yet more. Before text goes to speech synthesis, it's cleaned: formatting tags are stripped so that only words and punctuation remain. This is necessary, otherwise the voice would read out fragments of code. But that cleaning has a direct consequence: the word you'd set in bold becomes an ordinary word again, read exactly like its neighbours.
In other words, typographic emphasis isn't rendered badly, it isn't rendered at all. This isn't a flaw specific to our engine, it's the nature of the exercise: what remains is the text, and text alone doesn't say that a word was bold. If you want to understand in depth what separates reading a text from speaking it, that's the whole subject of our page on French text to speech.
Capitals, the opposite trap
With capitals, the risk isn't disappearance, it's a misread. A word written in full caps for emphasis, like "it's FREE", can be taken by an engine for an acronym and spelled out letter by letter: "F, R, E, E". The voice switches from a word to an alphabet, and the effect is genuinely grating. The behaviour depends on how the tool tells a shouted word from an acronym, which is exactly the question our article on acronyms and abbreviations in audio deals with.
On our side, a common word written in capitals is recognised as a word and read normally, not spelled out. But no automatic rule is perfect: a rare word or a brand in caps can still trip things up. When a specific term has to be said a fixed way, the right answer isn't to hope, it's to enter it once and for all in the pronunciation lexicon, which applies across the whole site.
Italics, where meaning is really at stake
Italics are trickier, because in careful writing they often carry meaning, not just insistence. They mark a foreign phrase, the title of a work, or signal that a word is being used as a word ("the term liaison"). On the ear, all of that falls away. "He didn't say it, he wrote it" loses its backbone: the contrast between the two verbs rested entirely on the italics, and without them the sentence goes flat.
That's the real lesson in all this. The problem isn't technical, it's editorial. A sentence whose meaning rests on formatting is a fragile sentence, even for a hurried reader skimming the page. In audio, it breaks outright.
Writing so the emphasis survives
The fix fits in one line: put the weight in the words, not in their dressing. Where you'd have stressed something visually, rephrase so the emphasis comes from the construction. "He didn't say it, he wrote it" is saved by making it explicit: "He never said it out loud; he wrote it down, in black and white." The contrast now lives in the words, and it can be heard.
Two concrete habits. First, punctuation does carry to the ear: a comma, a full stop, a colon create real pauses, and a well-punctuated text keeps its relief when spoken. That's the subject of our article on punctuation and how a voice breathes. Second, reread aloud, or better, listen to the audio version of your own text. A passage that loses its meaning when heard is a passage that leaned on the visual, and it's often a passage that would read better for everyone if it were clarified.
What we don't do, and why
We could imagine turning bold into a shift in intonation, italics into an inflection. We don't, and it's a deliberate choice. Emphasis added automatically to every bold word would quickly produce a theatrical, tiring read, with stresses landing where the author didn't necessarily want them. We'd rather deliver a faithful reading of the actual text than an interpretation of its formatting. Quality, for us, is saying well what is written, not guessing what the typography was trying to suggest.
Give your articles a voice with WeDispatch
This blog is itself voiced by WeDispatch. Curious how it sounds on your content?
Book a demo