Even worse, many of the mistakes looked plausible at first glance. A wrong homonym or slightly incorrect phrase could completely alter the meaning of a paragraph while remaining difficult to spot during proofreading.
For many dyslexic students and adults, this created a frustrating paradox. The very people who most needed support with writing and transcription often found speech-to-text impossible to edit. Some abandoned it entirely after a few discouraging experiences.
But over the past several years, something important has changed.
Newer AI-assisted speech systems are dramatically better than earlier generations of dictation software. While no system is perfect, many people who once gave up on speech-to-text may be surprised at how usable it has become.
The improvement is not simply that computers hear words more accurately.
Increasingly, modern systems use context, language prediction, pause patterns, and large language models to infer meaning and sentence structure. Instead of merely converting sounds into isolated words, newer systems attempt to understand language more like a conversation.
This has led to major improvements in automatic punctuation, sentence flow, and contextual word prediction. Earlier systems often produced long unpunctuated blocks of text or inserted incorrect but believable substitutions that required exhausting proofreading. Newer systems increasingly recognize pauses and sentence boundaries automatically, creating drafts that sound far more natural and readable.
Many dyslexic students and adults have significantly stronger oral expression than written transcription speed. Teachers and parents often notice that a student can explain an idea brilliantly in conversation yet struggle to produce a short written paragraph. Adults describe similar experiences in the workplace: strong ideas during meetings, but difficulty translating those thoughts into written communication quickly and efficiently. The issue is often not intelligence or understanding. It is hidden cognitive load.
Writing requires many processes to occur simultaneously: generating ideas, organizing thoughts, spelling, punctuation, keyboarding, rereading, self-monitoring, and remembering what comes next. For dyslexic individuals, these lower-level demands can consume enormous mental energy. By the time the mechanics of writing are managed, the original idea may already feel fragmented or exhausted.
By freeing mental energy for thinking, less effort will be needed for transcribing thoughts into printed words, making more energy available for imagination, reflection, reasoning, storytelling, and organization
People may also find their own ways to work with speech-to-text; so may talk aloud to themselves or talk through their ideas while driving or walking. With AI programs, it’s also possible to ramble aloud ideas and have AI models like ChatGPT, Google Gemini, or Grok organize, critique, summarize, or expand on what you say.
A majority of dyslexic adults seem to prefer separating idea generation and brainstorming from editing and proofreading, so text to speech along with an AI platform can work especially well.
So if you haven’t tried speech-to-text in a while, it may be a good idea to try it again. The best part is it’s free on every major platform
Speech to Text Systems:
Apple (iPhone, iPad, Mac Dictate),
Microsoft (Windows and Word Dictate)
Google (Android and Google Docs).
