DYSLEXIC AI
Posts
Newsletter 315: The Voice Tool That Finally Gets It Right

Newsletter 315: The Voice Tool That Finally Gets It Right

🧠 Why Wispr Flow Changes Everything for Dyslexic Thinkers Who Think Faster Than They Type

Matt "Coach" Ivey
November 18, 2025

A Quick Note About How This Newsletter Was Written (And Why I Desperately Needed This Tool)

I'm dictating this to you right now using Wispr Flow.

Not using my keyboard. Not fighting autocorrect. Not stopping every few seconds because my fingers can't keep up with my brain.

Just talking. And watching my thoughts appear on screen as coherent sentences.

But here's why I'm particularly excited about this right now: my MacBook's F5 button stopped working after my last system update.

You know what my F5 button does? It's linked to my microphone for dictation. The button I've been relying on for voice input. The one I press dozens of times a day because I love using voice for everything and really don't like typing.

That button breaking absolutely slowed me down and made life more difficult.

Suddenly I couldn't easily dictate into ChatGPT. Couldn't voice-type my emails. Couldn't capture thoughts quickly when working in different apps. I'd gotten so used to thinking out loud that losing that capability felt like someone had tied my hands behind my back.

Enter Wispr Flow.

Now? I just hold down the Fn key and talk. That's it. It translates for me instantly, everywhere.

Even better—and this is what sold me—it works in places where the large language models don't have microphone or dictate buttons. You know how sometimes you're using the online version of a tool versus the app version, and features are different?

Like today when I was using Claude.ai in my browser. The online version didn't have any microphone or dictation options. Before Wispr Flow, that meant I was stuck typing.

Now? Hold Fn. Talk. Done.

For someone whose dyslexic brain thinks primarily in voice rather than text, this feels like magic.

But it's not magic. It's just finally—FINALLY—voice-to-text technology that works the way my brain works.

Let me show you why this matters.

What You'll Learn Today

What Wispr Flow actually is (and why it's different from every dictation tool you've tried)
How the voice-to-text features handle real conversation, not robotic speech
Why the voice-to-voice features transform how you use AI tools
How this reduces cognitive load for dyslexic and neurodivergent thinkers
My personal experience with broken Mac dictation and why Wispr Flow saved me
Practical applications for homeschooling, work, content creation
How this connects to everything we've discussed about cognitive partner AI
Why voice-first tools are the future of neurodivergent productivity

Reading Time: 12-14 minutes | Listening Time: 9-11 minutes if read aloud

The Problem We've All Been Living With

Here's what writing has looked like for me for most of my life:

I have a thought. A complete, fully-formed, usually brilliant thought (if I do say so myself).

I sit down at the keyboard to type it out.

By the time my fingers hit the keys, the thought has fragmented into three different directions. I'm now trying to remember which version I wanted to write while also fighting dyslexic letter reversals and autocorrect that thinks it knows better than I do what word I meant.

Halfway through the sentence, I've forgotten where I was going with it.

I delete everything and start over.

Repeat this process about seventeen times, and you understand why I've published almost 320 newsletters but each one feels like running a mental marathon.

Voice dictation was supposed to solve this.

And it did... sort of.

I've tried:

Dragon NaturallySpeaking (cumbersome, required training, felt like work)
Built-in Mac dictation (which stopped working when my F5 key died)
Google Docs voice typing (okay but limited)
Siri dictation (laughably bad for anything longer than a text message)
Various AI transcription tools (require recording, then editing, then...)

Each one helped marginally. None of them felt natural.

None of them let me think at the speed my brain actually works.

And when my Mac's F5 button—my dictation hotkey—stopped working after the last update? I realized just how dependent I'd become on voice input.

Going back to typing full-time felt like trying to run with weights strapped to my ankles.

Until Wispr Flow.

What Wispr Flow Actually Is (And Why It's Different)

Wispr Flow is system-wide dictation that works everywhere on your Mac.

Not in one specific app. Not requiring you to record and transcribe. Not making you pause and say "period" or "comma" like a robot.

Everywhere. Seamlessly. Naturally.

You hold down a hotkey (I use the Fn key, but you can customize it). You talk. It appears as text in whatever application you're using.

Gmail. Slack. Google Docs. Notion. Claude.ai (even the online version with no built-in microphone). ChatGPT. NotebookLM. Your newsletter platform. Your text editor. Anywhere you can type, you can now speak.

This was critical for me when my F5 dictation button died. I wasn't locked out of voice input anymore. Wispr Flow became my new universal voice interface, and honestly, it works better than what I had before.

But here's what makes it actually usable for a dyslexic brain:

It understands context. It knows when "there," "their," and "they're" are correct based on what you're saying. It figures out punctuation from how you pause and speak. It handles complex sentences without you having to dictate formatting.

It keeps up with your thinking speed. I can ramble through a complete paragraph without it cutting me off or losing the thread. My brain thinks in paragraphs and full ideas, not sentence fragments. Wispr Flow accommodates that.

It learns how you speak. The more you use it, the better it gets at understanding your particular speech patterns, your vocabulary, your thinking style.

It requires zero setup or training. You install it. You customize your hotkey. You start using it. That's it.

For someone whose dyslexic brain operates at one speed verbally and a completely different (much slower) speed in typing, this is transformative.

The Voice-to-Text Features That Actually Work

Let me break down what makes Wispr Flow's voice-to-text different from everything else I've tried:

Natural Speech Patterns

I don't have to speak like I'm dictating to a court stenographer. I can talk like I'm having a conversation with you.

"So here's the thing about dyslexic minds and AI tools—we've been adapting to systems that don't match how we think for our entire lives, right? And now with AI, suddenly the system can adapt to us instead."

That's how I think. That's how I speak. And Wispr Flow captures it without me having to artificially structure it as:

"So here's the thing period about dyslexic minds and AI tools dash we've been adapting comma..."

The punctuation appears correctly based on how I spoke. The em dash shows up where I naturally paused with that slight tonal shift that indicates a related but separate thought. The question tag at the end gets its question mark.

It understands prosody—the rhythm and intonation of natural speech.

Context-Aware Corrections

When I say "their" in "their cognitive advantages," Wispr Flow knows I don't mean "there" or "they're" because of context. When I say "two" in "two years of research," it knows I don't mean "to" or "too."

This seems small until you realize how much cognitive energy you've been spending on fixing these errors in traditional dictation.

Every time the system gets a homophone wrong, you have to:

Notice the error
Stop speaking
Go back and fix it
Resume your train of thought
Hope you remember where you were going

Wispr Flow eliminates 90% of those interruptions.

Smart Formatting

Technical terms. Proper nouns. Specialized vocabulary. Company names.

Wispr Flow handles them surprisingly well, especially as it learns what words you use frequently.

I can say "ChatGPT" or "NotebookLM" or "dyslexic" or "neurodivergent" or "Cognitive Load Reduction" and it capitalizes, formats, and spells them correctly.

For content creators, educators, professionals—anyone who uses domain-specific language regularly—this is huge.

Speed and Responsiveness

There's essentially no lag between when I speak and when text appears.

This matters more than you might think. When there's a delay, you start speaking more slowly to compensate. Slower speech means slower thinking. You lose the flow state.

Wispr Flow keeps up with thinking speed, not typing speed.

Works Where Built-In Dictation Doesn't

This is what saved me when my F5 button died.

Built-in Mac dictation (when it was working) only functioned in certain apps or contexts. Some web applications wouldn't recognize it. Some text fields wouldn't accept it.

Wispr Flow works everywhere. Literally everywhere you can place a cursor and type, you can use Wispr Flow.

That online version of Claude.ai I mentioned? No microphone button in the browser interface. Before Wispr Flow, that meant typing only. Now I just hold Fn and talk directly into the chat window.

Same with any web app, any text field, any interface. If you can type there, you can speak there with Wispr Flow.

The Voice-to-Voice Features (This Is Where It Gets Really Good)

Voice-to-text is valuable. But voice-to-voice? That's transformative.

Here's what I mean: Wispr Flow doesn't just turn your speech into text. It can act as the interface between you and AI systems that also respond with voice.

With ChatGPT Voice Mode

When I'm using ChatGPT's voice mode, I used to have this awkward workflow:

Speak my question to ChatGPT
Listen to its voice response
If I wanted to reference something specific from its response, I'd have to manually type it or try to speak it clearly enough that ChatGPT understood what I was referencing

Now with Wispr Flow:

I speak my question naturally (Wispr Flow handles the input)
ChatGPT responds with voice
I can immediately follow up conversationally, and Wispr Flow captures my exact wording
The conversation flows like talking to a person, not interacting with a machine

With NotebookLM

This is where it gets really interesting for research and learning.

NotebookLM (Google's AI research tool that can analyze documents and have conversations about them) has a fantastic voice feature. You can have AI-generated discussions about your sources.

But taking notes during those discussions used to require:

Listening to the AI conversation
Pausing to type notes
Missing parts of the discussion while typing
Losing the flow of insight

With Wispr Flow:

I listen to the NotebookLM discussion
When I have a thought or want to capture something, I speak it
My notes appear in my document while the AI continues
I'm not pulled out of the learning experience to fight with a keyboard

It's the difference between being an active participant versus a frantic note-taker.

With Claude.ai (And Other Tools Without Built-In Voice)

Here's where Wispr Flow really shines: it brings voice capability to tools that don't have it.

The online version of Claude.ai? No microphone button. But with Wispr Flow, I can have voice conversations anyway. I hold Fn, speak my prompt, release, and Claude responds. Then I hold Fn again for my follow-up.

It's not quite as seamless as built-in voice-to-voice, but it's infinitely better than typing when your brain thinks in speech.

And as more AI tools add voice capabilities, Wispr Flow becomes the universal interface that works everywhere, whether the tool has native voice support or not.

Why This Matters for Neurodivergent Minds (Cognitive Load Reduction in Action)

Remember Newsletter 312 where I described Cognitive Load Reduction (CLR) as one of our core frameworks?

Wispr Flow is CLR incarnate.

The Cognitive Burden of Translation

For dyslexic brains, there's a translation step between thought and written word that most people don't experience.

My thoughts exist primarily as verbal-spatial concepts. Converting them to written text requires:

Formulating the verbal version (easy)
Converting verbal to written format (hard)
Handling motor control for typing (additional load)
Monitoring for dyslexic errors (more load)
Maintaining the original thought while doing all this (nearly impossible)

Each of these steps consumes working memory. By the time I get to step 5, I've often lost the original thought entirely.

Wispr Flow eliminates steps 2, 3, and 4.

I formulate the verbal version. I speak it. It appears correctly formatted as text. Done.

That's massive cognitive load reduction. All that mental energy I was spending on translation mechanics? Now available for actual thinking.

The Executive Function Angle

Many dyslexic people also have executive function challenges (hello, ADHD overlap).

Executive function involves:

Initiating tasks
Maintaining focus
Switching between subtasks
Completing work

Writing traditionally hits all of these:

"I should start writing" (initiation barrier)
"I need to focus on typing this correctly" (sustained attention)
"Wait, did I spell that right? Let me check" (task switching)
"Ugh, this is taking forever, I'll finish later" (completion failure)

With voice:

"I'll just talk about this" (lower initiation barrier)
"I'm just having a conversation" (easier to maintain focus)
"The typing is handled, I can stay in my thoughts" (no task switching)
"Oh, I'm done already?" (completion happens naturally)

Voice reduces executive function load by making the process feel like something we're already good at: talking.

The Confidence Factor

There's an emotional component too.

When writing is a constant battle with errors and slowness, you start dreading it. That dread becomes another form of cognitive load—anticipatory anxiety that makes everything harder.

With Wispr Flow, writing stops being a battle. It becomes natural expression.

That shift in how you feel about creating content? That's cognitive load reduction too. Less anxiety means more mental resources available for actual creativity and thinking.

When Your Tools Break (A Real-World Test)

Here's the thing about cognitive load: you don't always notice it until something changes.

When my F5 dictation button stopped working, I suddenly felt that load return. Every email took longer. Every thought I wanted to capture required switching to keyboard mode. Every newsletter idea that popped into my head while I was away from my desk got lost because I couldn't easily voice-capture it.

I realized just how much cognitive burden voice input had been removing.

Wispr Flow gave that back immediately. Same workflow. Actually better workflow. Lower barrier. More flexibility.

That's a real-world validation of cognitive load reduction.

How This Connects to Cognitive Partner AI (The Bigger Picture)

Wispr Flow isn't just a dictation tool. It's a piece of the cognitive partner AI ecosystem we've been building toward.

The Learn Your Way Connection

In Newsletter 313, we explored Google's Learn Your Way—AI-augmented textbooks that adapt to individual learners.

One of the key features? Audio lessons and narrated content for learners who process verbal information better than text.

Wispr Flow is the input side of that equation. Learn Your Way handles output (presenting information verbally). Wispr Flow handles input (capturing thoughts verbally).

Together, they represent a future where text-based interfaces are optional, not mandatory.

For dyslexic students who need to:

Listen to lessons (Learn Your Way)
Respond to questions verbally (Wispr Flow)
Take notes by speaking (Wispr Flow)
Interact with AI tutors conversationally (both)

The combination removes text as a barrier to learning entirely.

The 10-80-10 Rule Application

Remember our framework: dyslexic minds excel at idea generation (10%) and final polish (10%), while AI handles middle execution (80%).

Wispr Flow optimizes that first 10%—the idea generation phase where you're thinking rapidly, making connections, generating insights.

Without Wispr Flow: I have ideas → I try to type them → I lose half of them in translation

With Wispr Flow: I have ideas → I speak them → They're captured perfectly → I can generate the next idea

The 80% (organizing, formatting, structuring) can still be handled by AI. But now the initial 10% isn't bottlenecked by typing speed or dyslexic translation struggles.

Building Your Cognitive Partner Stack

We're assembling the pieces:

AI Models (ChatGPT, Claude, NotebookLM): The thinking partners
Personalized Prompts (dyslexic.ai library): The interaction framework
Adaptive Content (Learn Your Way): The learning materials
Voice Interface (Wispr Flow): The natural communication method

Each piece reduces cognitive load in a different way. Together, they create an environment where neurodivergent minds can operate at full capacity.

This is what cognitive partnership looks like in practice.

Practical Applications (How I'm Actually Using This)

Let me show you real-world scenarios:

Newsletter Writing

Before Wispr Flow (and after F5 broke):

Open Google Docs
Stare at blank page
Try to type introduction
Delete and restart multiple times
Take break out of frustration
Return and force through a draft
Spend twice as long editing for dyslexic errors
Publish exhausted

With Wispr Flow:

Open Google Docs
Hold Fn
Talk through the newsletter like I'm telling you about it
Watch it appear as coherent text
Do light editing for structure
Publish energized

Time saved: About 50%. Energy saved: Immeasurable.

Homeschooling with My Kids

Lesson Planning: I can think out loud about what we'll cover, creating lesson plans by describing them conversationally rather than typing structured documents.

Student Work: My daughter can respond to assignments verbally. Wispr Flow captures her thoughts without spelling being a barrier to demonstrating understanding.

Research Notes: When we're learning together and come across something interesting, we can capture notes by talking about it rather than stopping to type.

Working with AI When Native Voice Isn't Available

The Claude.ai Example: I was working in the online browser version today. No microphone button. Before Wispr Flow, that meant I was stuck typing long prompts and follow-up questions.

With Wispr Flow:

Hold Fn
Speak my complex prompt naturally
Release
Claude responds
Hold Fn again for follow-up

The conversation flows naturally even though Claude's browser version doesn't have native voice input.

This works for any AI tool, any web app, any text interface.

Client Coaching Calls

Before Wispr Flow:

Take rough notes during call
Spend time after call typing up detailed notes
Lose some nuance in the delay

With Wispr Flow:

Speak detailed notes immediately after key points
Capture exact wording while it's fresh
Have complete records without the typing delay

Content Creation

Social Media: Instead of drafting posts in notes then copying to platform, I just speak directly into the social media post window.

Email Responses: Long, thoughtful emails get done in minutes instead of being on my "I should reply to that" list for days.

Documentation: Technical documentation or processes I need to record? I just explain them out loud.

The Voice-First Future (Why This Matters Beyond One Tool)

Wispr Flow is one implementation of a broader trend: interfaces are becoming voice-native rather than text-native.

Why This Transformation Matters

For centuries, text literacy was the only gateway to knowledge work. If you couldn't read and write proficiently, you were excluded from most professional opportunities.

That created massive barriers for:

Dyslexic individuals
People with visual impairments
Those with motor control challenges
Non-native speakers in text-heavy environments
Anyone whose primary processing mode isn't textual

Voice-first interfaces demolish those barriers.

The Neurodivergent Advantage (Again)

Here's something interesting: neurodivergent people who've been forced to develop alternative strategies may have an advantage as interfaces go voice-first.

We're already comfortable:

Thinking verbally
Processing information aurally
Communicating conversationally
Working with AI as cognitive partners (we've been doing cognitive partnership our whole lives through accommodations and workarounds)

As the world shifts from text-primary to voice-optional, the skills we developed out of necessity become mainstream advantages.

We're not catching up anymore. We're ahead.

What Comes Next

Wispr Flow is excellent, but it's early days for voice-first computing.

Imagine:

Voice-native operating systems
AI assistants that think with you conversationally
Educational systems where text is one option among many
Professional environments where verbal processing is valued equally to written

That future isn't distant. It's being built right now.

And tools like Wispr Flow are showing what's possible when we stop trying to force square pegs (verbal thinkers) into round holes (text-required systems).

How to Get Started (Try It Yourself)

I'm genuinely excited about Wispr Flow in a way I haven't been about a productivity tool in years.

Here's how to try it: https://wisprflow.ai/r?MATT1424

It works on Mac (sorry PC users, not available yet). Installation takes minutes. Setup is minimal. You can start using it immediately.

A few tips from my experience:

Start small: Use it for emails or short documents first. Get comfortable with how it handles your speech patterns.

Customize your hotkey: I use Fn, but you can choose whatever works for you. Just pick something comfortable that you can hold while speaking.

Find your rhythm: Some people prefer speaking in short bursts. Others (like me) prefer full paragraph-style thinking. Both work.

Don't over-correct: Trust it to get punctuation and formatting right. The more you interrupt yourself to manually fix things, the less natural it feels.

Use it with AI: The voice-to-voice capabilities with ChatGPT or NotebookLM are where it really shines. But even with tools that don't have native voice (like Claude.ai browser version), it works great.

Pair it with our prompts: The 90+ prompts at dyslexic.ai work even better when you can speak them naturally instead of typing them.

Test it everywhere: Try it in different apps, web interfaces, text fields. You'll be surprised where it works seamlessly.

A Personal Note on Tools vs. Systems

I'm careful about recommending specific tools in these newsletters.

Tools come and go. Companies change. Products get abandoned or acquired.

What matters more than any specific tool are the principles and systems that make neurodivergent cognition work better.

But Wispr Flow exemplifies those principles so clearly—and solved such an immediate pain point for me when my built-in dictation broke—that it's worth highlighting:

✓ Reduces cognitive load by eliminating translation steps ✓ Works with natural thinking patterns rather than forcing artificial structure ✓ Enables cognitive partnership by being the interface between you and AI ✓ Removes barriers that have nothing to do with intelligence or capability ✓ Feels effortless rather than requiring constant willpower ✓ Works universally across all apps and interfaces

Even if Wispr Flow disappears tomorrow (which I doubt), the principle remains: voice-first interfaces are transformative for dyslexic and neurodivergent minds.

Whatever tools emerge next, look for these qualities.

And when your regular tools break (like my F5 button), have backups ready that are actually better than what you had before.

TL;DR - Too Long; Didn't Read For Fellow Skimmers: The Essential Points

🎤 What It Is: Wispr Flow is system-wide dictation that works everywhere on your Mac, naturally and seamlessly

💔 Why I Needed It: My Mac's F5 dictation button broke after an update—Wispr Flow saved my workflow

✨ Voice-to-Text: Understands context, natural speech patterns, formats automatically—just talk normally

🗣️ Voice-to-Voice: Works perfectly with ChatGPT, NotebookLM, and other AI voice interfaces

🌐 Works Everywhere: Even in tools without native voice (like Claude.ai browser version)

🧠 Cognitive Load Reduction: Eliminates translation steps between verbal thought and written text

⚡ For Dyslexic Minds: Think at verbal speed, skip typing struggles, reduce executive function burden

🎯 Cognitive Partner Connection: The input side of adaptive learning (pairs with Learn Your Way's output)

📚 Practical Uses: Newsletter writing, homeschooling, client notes, content creation, email, AI conversations

🚀 The Future: Voice-first interfaces are the future; neurodivergent thinkers are already prepared

💡 How to Try: https://wisprflow.ai/r?MATT1424 (Mac only, minimal setup)

🔧 Best Practices: Hold Fn (or your hotkey), talk naturally, trust the system, use everywhere

Bottom Line: This is the voice tool that finally works the way dyslexic brains actually think. And it saved me when my regular dictation broke.

Have a Great Week—And Try Thinking Out Loud

This entire newsletter was written by talking.

Not dictated robotically. Not painstakingly transcribed and edited. Just... talked.

The way I think. The way I process. The way I naturally communicate.

And you're reading it as coherent, structured, formatted text.

That's the magic Wispr Flow enables.

For dyslexic minds who've spent our lives fighting with keyboards—and then fighting with broken dictation buttons—this feels revolutionary.

So this week, try it. Download Wispr Flow at https://wisprflow.ai/r?MATT1424. Hold down Fn (or whatever hotkey you choose). And just start talking.

Tell the AI what you're working on. Dictate your emails. Speak your thoughts into your notes. Have a voice conversation with ChatGPT or NotebookLM about a problem you're solving. Use it in Claude.ai even though the browser version doesn't have a microphone button.

Experience what it feels like when technology adapts to how your brain works rather than forcing you to adapt to it.

That's cognitive partnership. That's the future we're building. That's what tools like this make possible.

And if you discover, like I have, that you think better when you can speak your thoughts rather than type them?

Welcome to the voice-first future. We've been waiting for you.

— Matt "Coach" Ivey, Founder · LM Lab AI

(Dictated, not typed. Because my F5 button is still broken, but I don't need it anymore.)

Take Action This Week

Try Wispr Flow: https://wisprflow.ai/r?MATT1424 - Download for Mac, start using immediately

Experiment with Voice-to-Voice: Use it with ChatGPT voice mode or NotebookLM audio features

Test in "No Voice" Environments: Try it in Claude.ai browser, web apps, any interface without native voice input

Apply Our Prompts: Use the 90+ prompts at dyslexic.ai but speak them instead of typing them

Share Your Experience: Reply to this email—did voice-first tools change your workflow?

Explore Cognitive Partnership: Read Newsletter 313 (Learn Your Way) and Newsletter 312 (our frameworks)

Join Our Community: Discuss voice-first strategies and tools at dyslexic.ai

Book a Call: Need help setting up your voice-first cognitive partner system? Schedule on our site

Tell Others: Share this with anyone who thinks faster than they type (or has broken dictation buttons)