My name is Roni, I am a serial solo builder & founder based in Stockholm 🇸🇪
I built wisprkey to be more productive in my AI coding workflow, prompting Claude Code/Codex was the bottleneck. Every voice-typing tool I tried had a limited free tier or a pricy pro plan, so I built one that is 100% free.
What’s in it: 100% Free to use: unlimited voice typing, 31 languages, 98% accuracy. Pro ($4.99/mo): text-to-speech in 64 voices, translation to 31 languages. Agent cursor coming soon.
Appreciate feedback 🙏
// Roni
Report
congrats on the launch. the "vibe coding" tag next to a dictation app caught my eye - is accuracy tuned at all for spoken code/technical terms (variable names, "snake case", "curly brace", etc.) or is it more general-purpose dictation that happens to work fine for prompting an AI assistant in plain English? asking because that's usually where general dictation tools fall apart for devs.
@galdayan My workflow: prompt in natural language, then listen to the output.
This is the only way I can run 10 agentic loops at the same time in CC/Codex.
Technical term would be a great new feature, thanks for the feedback!
Report
I run my whole business through AI agent sessions now, mostly long, rambling prompts where I'm thinking out loud about context before I get to the actual ask. Does accuracy hold up on that kind of multi-minute dictation with pauses and false starts, or is it tuned more for short, clean commands? The 100% free unlimited tier is the right call too — pricing is usually what pushes people back to typing on the tools that could otherwise replace it.
Report
I like that this is focused on removing friction rather than adding more AI features. When you're bouncing between coding, prompting, and writing, even small interruptions add up. A global voice shortcut that's always available feels like one of those tools you stop thinking about because it becomes part of your workflow.
Report
After recording, the correct text can be displayed on the FLOW interface, but if it is placed in the application scene, the text or picture of the pasteboard will be displayed
Wisprkey
Hey 👋
My name is Roni, I am a serial solo builder & founder based in Stockholm 🇸🇪
I built wisprkey to be more productive in my AI coding workflow, prompting Claude Code/Codex was the bottleneck. Every voice-typing tool I tried had a limited free tier or a pricy pro plan, so I built one that is 100% free.
What’s in it:
100% Free to use: unlimited voice typing, 31 languages, 98% accuracy.
Pro ($4.99/mo): text-to-speech in 64 voices, translation to 31 languages. Agent cursor coming soon.
Appreciate feedback 🙏
// Roni
congrats on the launch. the "vibe coding" tag next to a dictation app caught my eye - is accuracy tuned at all for spoken code/technical terms (variable names, "snake case", "curly brace", etc.) or is it more general-purpose dictation that happens to work fine for prompting an AI assistant in plain English? asking because that's usually where general dictation tools fall apart for devs.
Wisprkey
@galdayan My workflow: prompt in natural language, then listen to the output.
This is the only way I can run 10 agentic loops at the same time in CC/Codex.
Technical term would be a great new feature, thanks for the feedback!
I run my whole business through AI agent sessions now, mostly long, rambling prompts where I'm thinking out loud about context before I get to the actual ask. Does accuracy hold up on that kind of multi-minute dictation with pauses and false starts, or is it tuned more for short, clean commands? The 100% free unlimited tier is the right call too — pricing is usually what pushes people back to typing on the tools that could otherwise replace it.
I like that this is focused on removing friction rather than adding more AI features. When you're bouncing between coding, prompting, and writing, even small interruptions add up. A global voice shortcut that's always available feels like one of those tools you stop thinking about because it becomes part of your workflow.
After recording, the correct text can be displayed on the FLOW interface, but if it is placed in the application scene, the text or picture of the pasteboard will be displayed
Wisprkey
New users will be upgraded to Pro for free.
Tired of reading agent output? Listen to it.
我讲中文，它总是翻译成别的语言
@justin_jiaa Which model have you selected?