How to listen to any web article aloud, cleanly

A web page is not an article. It is an article wrapped in a navigation bar, a cookie banner, three ad slots, a newsletter popup, a row of social share buttons, a comment section, and a footer full of links to other articles. When you point a text-to-speech tool at the whole page, it reads all of that too. You get "Accept all cookies. Subscribe to our newsletter. Menu. Home. About. Contact" before it ever reaches the first sentence you actually wanted to hear.

The fix has a name: readability extraction. It is the process of looking at a page's HTML and pulling out just the main body text — the headline and the paragraphs — while discarding the chrome around it. This is the exact same idea behind Safari Reader and Edge's Immersive Reader. When you tap that little reader icon and the page collapses into clean text on a plain background, an extraction algorithm just decided which parts of the page were the article and which parts were noise. Text-to-speech that reads well starts from the same step.

There are two practical routes to a clean listen. The first is to let your browser do the extraction for you. Open the article, switch on your browser's built-in reader mode — Reader in Safari, Immersive Reader in Edge, Reader View in Firefox — and then start reading from that stripped-down version. Because the page is now just text, whatever reads it aloud has nothing left to trip over. The second route is to hand the article text to a reading tool directly: copy the article body and paste it in, or paste the URL and let the tool fetch and extract the main content for you. Wordcast is one place to paste text and press Listen without a signup.

Here is the honest limit, and it is worth understanding before you expect too much. A tool that runs entirely in your browser cannot always fetch an arbitrary URL's HTML. The reason is CORS — cross-origin resource sharing. When a web page tries to fetch content from a different site, the browser blocks that request unless the other site explicitly sends a header saying "anyone may read me." Most news sites and blogs do not send that header, because they never intended other pages to pull their content. So a purely client-side reader often cannot load the URL you paste. To read from a URL reliably, a tool needs a small server or proxy that fetches the page on its behalf, outside the browser's cross-origin rules. When there is no server in the loop, the dependable path is the one that never touches CORS at all: you paste the text yourself.

Extraction has its own failure cases, too, and they are not bugs you can wish away. Paywalled articles frequently do not extract cleanly, because the full text is not in the page you can reach — it is behind a login, or delivered only after payment, and what extraction sees is the teaser. Pages that build their content with JavaScript after loading can also come back nearly empty to a simple fetch, because the raw HTML holds a skeleton and the words arrive later from a script. In both cases the cleanest move is the manual one: open the article the normal way, select the text you can see, and paste that. Your eyes are a very good extraction algorithm.

Once you have clean text going in, the listening itself is where this pays off. Queue a long read while you are on your commute and let it play through without touching the screen. Start an article at the stove and keep cooking. Pick a natural-sounding voice rather than the first robotic one in the list — the difference over twenty minutes is the difference between finishing and giving up. And nudge the speed. Most people who read a lot by ear settle somewhere between 1.1x and 1.5x once their ear adjusts; start at normal speed and push it up a notch at a time until it stops feeling comfortable, then drop back one.

There is a quiet privacy benefit to doing this in the browser, too. If the voice reading your article is an on-device voice — the ones your operating system ships for accessibility, with no "Google" or cloud label — then the article text never leaves your machine to be spoken. It goes from the page, to your browser, to the local speech engine, and out your speakers. Nothing about what you are reading is uploaded anywhere. For a private email or a work document you are catching up on, that is the difference that matters.

So the whole workflow is short. Get the article down to just its text, either by turning on your browser's reader mode first or by pasting the text into a reading tool. Expect URL fetching to sometimes fail on CORS, paywalls, or script-rendered pages, and fall back to pasting when it does. Then pick a good voice, set a speed that suits your ear, and go do something else while you listen.