# Text to Speech Accessibility: What to Check Before You Pick URL: https://heartheweb.com/journal/text-to-speech-accessibility Type: blog Locale: en Published: 2026-10-15 Updated: 2026-10-15 --- > TTS is not a screen reader. Five checks for low vision, dyslexia and hands-free reading, plus four apps compared on speed range, highlighting and price. Text to speech accessibility comes down to five checks: speed range, highlighting, resume behavior, screen reader coexistence and format fidelity. In short, does the voice stay clear at your listening speed, and does the app work with the screen reader you already use? A TTS app is not a screen reader, and treating it as one is the most common mistake. Below, the difference, the checklist, and where each tool actually holds up. ## Is text to speech the same thing as a screen reader? No, and the confusion costs people months. A screen reader (VoiceOver, TalkBack, NVDA, JAWS) narrates the interface itself: buttons, menus, form fields, the role of every element. You drive it with gestures or keys, and it replaces the screen for people who cannot use one. Text to speech reads a document you hand it. It is a read-only layer. It will not buy anything, fill a form or tell you which button is focused. What it gives you is a clean, continuous narration of an article, PDF or book, usually with a better voice and a better speed control than the screen reader's default. The practical split: a screen reader is how you operate the device. TTS is how you consume long text on it. Many low vision readers use both on the same day, sometimes on the same page. === SCALE CHECK === The [World Health Organization estimates](https://www.who.int/news-room/fact-sheets/detail/blindness-and-visual-impairment) that at least 2.2 billion people have a near or distance vision impairment. Most of them are not blind. They are people who can see a screen but not comfortably for 40 minutes, and for them a listening queue is a reading tool, not a novelty. ![Hand holding a phone on a commuter train with earbuds plugged in](https://fdzlnqpwsaniezitwiuw.supabase.co/storage/v1/object/public/cms-media/heartheweb/2026-10/c1436b-commute.webp) ## What does a text to speech accessibility checklist look like? Five checks, in order of how often they break. Run them with a real article before you pay for anything. - **Speed range.** Default narration sits around 150 words per minute. Most readers silently read at roughly 240. If you listen to learn, you will push to 1.5x or 2x within a week, so the voice has to stay intelligible there. Voice Dream Reader lists a range of 50 to 500 WPM, which is the widest we have seen documented. - **Synchronized highlighting.** Word or sentence highlighting keeps your eyes anchored if you have partial sight, dyslexia, or attention that drifts. NaturalReader and Voice Dream both do it well. - **Resume behavior.** Take a phone call, come back. Does playback restart at the top of a 6,000-word piece or at the paragraph you left? Readwise Reader starts from the paragraph centered on screen, which is the right idea. - **Screen reader coexistence.** If VoiceOver or TalkBack is on, does the TTS app's player still expose labelled controls, or does it trap focus in a canvas? This is the check that most vendor pages skip. - **Format fidelity.** Footnotes, nested quotes, tables, image alt text. Does the narrator read "footnote 14" mid-sentence, skip the table, or announce every link URL? ## Do screen reader users even need a separate TTS app? Often yes, for one reason: the screen reader is built to operate, not to narrate. It stops at every element, announces roles, and reads long text in a flat, utilitarian voice tuned for speed of navigation. The [WebAIM Screen Reader User Survey #10](https://webaim.org/projects/screenreadersurvey10/) found that 91.3% of respondents use a screen reader on a mobile device. That is the context where a long article is most awkward: small screen, one thumb, a swipe per paragraph. Handing the article to a dedicated narrator, then pressing play, removes hundreds of swipes. Skip the separate app if all you read is short text and the built-in voice is already fine for you. Pay for one if you read long-form, PDFs or newsletters every day. ![Finger swiping a smartphone screen with headphones around the neck](https://fdzlnqpwsaniezitwiuw.supabase.co/storage/v1/object/public/cms-media/heartheweb/2026-10/282d91-swipe.webp) ## Which TTS apps hold up for low vision, dyslexia and hands-free reading? Four we would put in front of someone starting from zero. Prices are from each vendor's pricing page at the time of writing and move often. **Voice Dream Reader** is the accessibility-first pick. It started as a tool for blind and low vision readers and still has the deepest VoiceOver integration in the category. It handles PDF, EPUB, DAISY with real chapter navigation, Word and plain text. The catch: Apple only, and about $79.99 a year on iPhone and iPad after a 7-day trial. **NaturalReader** is the cheapest way to test the idea. The free tier has unlimited basic voices, with premium voices capped at roughly 20 minutes a day. Plus runs $20.90 a month or $119 a year. Sentence highlighting, dual PDF views (original layout and reflowed) and camera OCR make it strong for dyslexia and for printed material. **Speechify** has the larger app ecosystem and the more polished mobile player. Premium is $29 a month or $139 a year, with speeds up to 4.5x. It is built for consumption at volume. Its weak spot for this audience is price: the free tier caps speed at 1.5x, so the speeds low vision readers tend to want sit behind the paid plan. **Readwise Reader** is a read-it-later app first and a narrator second. If your queue already lives there, its mobile TTS is good enough, and you keep highlights tied to what you heard. If you need PDF-heavy or DAISY workflows, look elsewhere. ## Why does narration quality matter more than the feature list? Because listening effort is the real cost. A voice that is slightly off on stress and pauses is fine for a 3-minute email and exhausting at minute 25 of an essay. People with low vision or dyslexia are not using TTS for fun at that length, they are using it to get through the work or the reading they would otherwise skip. Test with a 3,000-word piece that has subheads, a block quote and a list. Listen for three things: how the voice handles a heading versus body text, whether it breathes at paragraph breaks, and whether it mangles names. A narrator that reads "Dr." as "drive" will cost you trust in the whole queue. ![Tablet with enlarged text, headphones and reading glasses on a desk](https://fdzlnqpwsaniezitwiuw.supabase.co/storage/v1/object/public/cms-media/heartheweb/2026-10/34f67d-largeprint.webp) ### How fast should you set the speed without losing the thread? Raise it in small steps, not in one jump. Go from 1.0x to 1.25x for a few days, then 1.5x. Comprehension drops first on dense material (legal text, a paper with numbers) and last on narrative prose, so keep two speeds: a fast one for familiar topics and a slower one for anything you will need to act on. Experienced screen reader users often run synthetic voices far faster than a sighted listener would tolerate, and that is a skill built over years, not a setting to copy. Pick the speed where you could repeat the last sentence back. If you cannot, slow down by one notch. Also check what the app does with pitch. Some players speed up by shifting pitch, which makes voices thin and tiring. Good ones time-stretch and keep the voice stable. ### Which formats still break most narrators? Four formats cause most of the trouble we see, whatever the app: - **Scanned PDFs.** No text layer means no narration. You need OCR first, and NaturalReader's camera OCR or a separate scanner app does that job. - **Tables.** Most narrators read cell by cell with no column context. Expect to skim the original or ask for a summary. - **Math and code.** Equations and code blocks come out as noise unless the app has a dedicated mode. - **Paywalled and logged-in pages.** A browser extension sees what you see. A server-side converter often does not, so test one of your own paywalled sources before relying on it. None of these is a reason to skip TTS. They are the places to verify with your own documents during the trial week, and the reason the five-point checklist ends with format fidelity. ## What about the free options built into your phone and browser? Use them first. They are better than their reputation and cost nothing. - **iPhone and iPad:** Spoken Content, with Speak Screen (two-finger swipe down) and Speak Selection. - **Android:** Select to Speak, plus TalkBack if you need full navigation. - **Mac:** Speak Selection and the system voices. - **Edge:** the Read Aloud command reads web pages with natural voices and highlights as it goes. Where they fall short: no queue, no private feed, weak PDF handling, and no way to pick up an article on another device. If you are curious how browser tools compare, we covered the trade-offs in [read aloud extensions](/journal/read-aloud-extension). ## Does text to speech accessibility stop at the app you choose? No. The source matters as much as the narrator. A page with a proper heading structure, real lists and alt text on images produces a decent narration in any tool. A page built from div soup, with text baked into images, produces a bad one in all of them. This is also why a "listen to this page" widget on a site is no substitute for accessible markup. It reads the page for people who can already use the page. It does not help the reader whose screen reader cannot find the main content in the first place. If you publish, fix the structure first. If you read, pick tools that extract the article cleanly before narrating it. For study material in particular, see our notes on [text to speech for studying](/journal/text-to-speech-for-studying). ## Before your next commute `QUEUE: 3 articles VOICE: narrator-calm SPEED: 1.4x -> test at 1.8x STATUS: READY TO TRANSMIT`Start with what is already on your phone. Load one long article, set the speed where you can follow without effort, and note where it breaks: a footnote, a table, a lost place after a call. That list of breakages is your real shopping list. If the built-in voice is enough, stop there. If you hit two or more of the five checks above, try Voice Dream Reader on Apple devices or the NaturalReader free tier elsewhere, then decide. And if your reading queue is mostly web articles, a private MP3 feed you can open in any podcast app removes the player question altogether. That is what we build at hear.the.web, and it is one option, not the only one. We are the hear.the.web team, cambridge, ma. Tell us which check broke for you. ## FAQ ### Is text to speech the same as a screen reader? No. A screen reader narrates the whole interface and lets you operate the device. Text to speech reads a document you choose, such as an article or PDF, and cannot perform actions like filling a form. ### What is the best text to speech app for low vision? Voice Dream Reader has the deepest VoiceOver integration and a 50 to 500 WPM speed range, but it is Apple only at about $79.99 a year. NaturalReader is a cheaper cross-platform start with a usable free tier. ### Can text to speech help with dyslexia? Many readers with dyslexia use TTS with synchronized word highlighting, which keeps the eyes anchored while the narrator reads. NaturalReader and Voice Dream Reader both offer highlighting. ### Are the built-in TTS tools on iPhone and Android good enough? For short text, yes. Spoken Content on iPhone and Select to Speak on Android are free. They lack queues, private feeds and strong PDF handling, which is where paid apps earn their price. ### What speed should I listen at? Start at 1.0x and move up in small steps, such as 1.25x then 1.5x. Choose the speed where you could repeat the last sentence back, and keep a slower setting for dense material. ### Does a read-aloud widget make a website accessible? No. A widget narrates a page for people who can already use it. Accessibility depends on real heading structure, alt text and labelled controls that screen readers can navigate.