For more detail, see our analysis of how to read federal reserve meeting minutes.
The Workflowy note that started this was blunt: “Fast blog writing – voice recognition.” I’d written it and ignored it for three months, because the idea seemed like it would create more work than it saved. For more detail, see this deep-dive on what is containerization? docker explained for people who don’t write code.
I’ve spent a lot of time researching this topic, and here’s what I found.
Here’s the thing most people miss about this topic.
It doesn’t. Here’s what I actually do.
Why Typing Is Slower Than You Think
The bottleneck in writing isn’t ideas — it’s the translation from thought to text. Typing at 60-80 words per minute is roughly one-third the speed of natural speech (120-180 words per minute for most people). But the more significant gap is cognitive: typing requires you to hold ideas in working memory while executing a motor sequence, which creates interference with the generative process [1]. For more detail, see our analysis of google docs vs microsoft word vs notion.
Related: digital note-taking guide
Speech bypasses this. When you’re talking, the motor system handles output automatically, leaving working memory fully available for content generation. The result is drafts that are less polished but more idea-dense — the kind of raw material that’s much faster to edit than to generate from scratch.
This is not a new observation. Academic research on voice transcription as a writing tool goes back decades, and the consistent finding is that speech-to-text produces first drafts at 2-3x the
The Actual Workflow
I use Samsung Galaxy’s built-in keyboard speech recognition (Samsung Keyboard → Microphone icon). It’s not the most accurate speech-to-text available, but it runs locally, requires no additional apps, and is available instantly in any text input field. For Korean, accuracy is excellent. For English, it’s good enough for a first draft — which is all I need.
Step-by-step:
- Open a Google Doc on mobile (or any text editor — I prefer Docs for cloud backup)
- Tap the microphone on the Samsung keyboard
- Talk through the post structure first: “This post is about X. The main sections are: Y, Z, W. The opening situation is…” (30 seconds, never included in final draft)
- Talk through each section without stopping to self-edit. If you lose the thread, say “delete that” and Samsung keyboard will undo. Don’t correct mid-sentence — it kills momentum.
- Add section headers by saying “new line” and typing them manually
A 1,000-word draft typically takes 8-12 minutes of actual talking time. The raw output looks like a transcript — run-on sentences, missing punctuation, some mis-recognitions. This is expected and fine.
The Cleanup Pass
I paste the raw transcript into Claude (or any capable LLM) with this prompt:
This is a voice-dictated blog draft. Clean up run-ons, add punctuation, fix speech-to-text errors. Preserve my voice — don’t rewrite sentences, just fix transcription artifacts. Return as plain text with section headers preserved.
The cleanup takes 30-60 seconds and produces something that requires one editing pass rather than a full rewrite [3].
Total workflow time for a 1,000-word post: 8-12 minutes dictation + 2-3 minutes cleanup + 15-20 minutes editing = roughly 30 minutes. Typed from scratch, the same post takes me 90-120 minutes.
Where It Doesn’t Work
This workflow fails for posts requiring precise technical language — anything with specific terminology that speech recognition consistently misconstrues. It also fails when I’m writing in a location where talking aloud is socially awkward, which is more of my writing contexts than I’d like.
The deeper limitation is that some writing benefits from the slower pace of typing — the constraint forces more precise thinking. Complex arguments with nested logic often work better typed. Narrative posts, practical guides, opinion pieces: voice wins. Technical analysis: type.
I’ve stopped treating voice dictation as a hack and started treating it as a different mode for different content types. The notebook I write in now has a column for “voice first” or “type first” next to each planned post topic.
Last updated: 2026-04-13
Your Next Steps
- Today: Pick one idea from this article and try it before bed tonight.
- This week: Track your results for 5 days — even a simple notes app works.
- Next 30 days: Review what worked, drop what didn’t, and build your personal system.
About the Author
Written by the Rational Growth editorial team. Our health and psychology content is informed by peer-reviewed research, clinical guidelines, and real-world experience. We follow strict editorial standards and cite primary sources throughout.
Disclaimer: This article is for educational and informational purposes only. It is not a substitute for professional medical advice, diagnosis, or treatment. Always consult a qualified healthcare provider with any questions about a medical condition.
Does this match your experience?
I believe this deserves more attention than it gets.
Key Takeaways and Action Steps
Use these practical steps to apply what you have learned about Voice-to-Blog:
Ever noticed this pattern in your own life?
References
- Variani, E. and Riley, M. (2025). Speech-to-Retrieval (S2R): A new approach to voice search. Google Research Blog. Link
- AssemblyAI Team (n.d.). The top free Speech-to-Text APIs, AI Models, and Open Source Engines. AssemblyAI Blog. Link
- Lermen, S. (n.d.). Run Local Speech-to-Text Transcription. Simon Lermen Substack. Link
- Sotto Team (n.d.). Blog – Voice-to-Text Tips & Guides. Sotto Blog. Link
- VoxWrite Team (n.d.). Dictate Blog Posts Faster: From Ideas to Published Content. VoxWrite Blog. Link
Frequently Asked Questions
What is the key takeaway about voice-to-blog?
Evidence-based approaches consistently outperform conventional wisdom. Start with the data, not assumptions, and give any strategy at least 30 days before judging results.
How should beginners approach voice-to-blog?
Pick one actionable insight from this guide and implement it today. Small, consistent actions compound faster than ambitious plans that never start.
Related Posts
- Open Source vs Proprietary Software: What the Difference Means for You
- Password Managers Explained: Why You Need One and Which to Choose
- How to Use Perplexity AI for Academic Research
