For testing YapClean with your own AI key.
- 100,000 initial output characters
- Then 5,000 free output characters per day
- OpenAI or Groq BYOK for speech
- Provider usage paid to your provider
- Windows beta access request
Windows Editorial Edition. Local Whisper or Fast Cloud.
Clean voice at your cursor.
Click a Windows text field, press CapsLock, speak naturally, and YapClean cleans the result before inserting it where you were already writing.
Designed for Windows users who value speed and total privacy. Setup is straightforward: bring your own OpenAI/Groq API key or run completely offline with downloadable local Whisper models.
Active Application Input
CapsLock listening
Notion
Listening
Press CapsLock again to insert
I am drafting our project requirements inside Notion and need to insert a paragraph about...
Cleaned & Inserted
the premium audio processing pipeline.
Cleaned Text Inserted
Autospace Active
Security First
Data Sovereignty: 100% Privacy
Your voice is your most personal data. YapClean is engineered around a zero-trust architecture to ensure that your spoken words, drafts, and text inputs never fall into the wrong hands.
What We Guarantee
- 100% Local Only Option — Local audio models process speech on your machine. Network is blocked.
- Auto-Sweep History — History is cleared automatically after your chosen duration (or instantly).
- No Server Logs — No centralized databases, audio caches, or telemetry profiles are kept.
What Competitors Do
- Stream your audio to cloud-based servers without permission.
- Save voice recordings to target future advertisements.
- Retain perpetual text histories linked to your identity.
Multi-Tier Architecture
Select Your Processing Engine
Whether you want completely offline dictation, blisteringly fast cloud responses, or managed convenience, YapClean adapts to your environment.
Run Completely Offline
Download and run optimized Whisper and Parakeet models natively on your Windows device. Keep your voice completely isolated from the internet with our strictly sandboxed Local-Only mode.
Model Footprint & RAM vs Accuracy
Whisper Tiny (74MB)
1.5 GB RAM
Whisper Small (465MB)
3.2 GB RAM
Whisper Large v3 (1.5GB)
6.0 GB RAM
Screamingly Fast Cloud API
Connect YapClean directly to your OpenAI or Groq API endpoints. Achieve instant transcription using ultra-low latency setups. Perfect for developers, technical founders, and operators.
Transcription Latency (Shorter is Better)
Groq API (Whisper V3)
0.32s
Standard API
1.28s
Managed Infrastructure
Prefer hands-free setup? Our Managed Cloud plans route transcription and cleaning pipelines securely through our optimized endpoints. No configuration keys required.
Under Development: Our team is completing the final limits, billing integrations, and automated server-side cost controls. Join our waitlist to be notified when managed registrations open!
Interactive Context
Smart Floating Widget
Our floating bouncing soundwave widget is designed to be elegant, fast, and completely aware of its surroundings.
It automatically detects active Windows text inputs (such as Telegram, Notion, Notepad, and Obsidian) to apply context-aware lowercase, adaptive spacing, and sentence-structure formatting directly at your cursor.
Floating Widget Status Preview
Soundwave Active
Formatting for Notion...
Watch the soundwave bars dynamically move based on real-time micro-inputs!
Bulk Upload Dashboard: /files
Ready to transcribe
Перетащите файлы сюда или выберите на компьютере
Supports high-speed local processing of audio and video records
.MP3
.WAV
.M4A
.FLAC
Bulk Transcription
Native Drag-and-Drop Audio
Transcribe meetings, lectures, files, and pre-recorded memos. Our dedicated /files tab integrates a native drag-and-drop landing area with complete support for major audio files.
Drop any recording, select your transcription model badge, and convert spoken words into crisp markdown paragraphs instantly.
Beta pricing
Start with Free BYOK. Upgrade only if the first loop works.
Manual access
BYOK provider costs separate
Windows only
Compatibility varies
Founder support
For BYOK users who want long-term YapClean access without a monthly YapClean subscription. Planned later lifetime price: $149.
- 100,000 initial trial characters
- 5,000 free daily BYOK characters afterward
- Polar is the selected checkout path
- YapClean license key issued by our backend
- No device-limit claim today
For users who want YapClean to manage model access when Managed Pro or Managed Max opens.
- Managed Pro draft: $19/month, 100,000 characters/month
- Managed Max draft: $49/month, 500,000 characters/month
- Not live until provider costs, limits, and QA are complete
Founder Lifetime BYOK is the first paid path. Managed plans are in development, and provider usage remains separate from the YapClean license.
Workflow
One key. One field. Clean output.
01
Click a Windows text field.
Start where you already plan to write: an email, chat box, note, form, CRM field, or document.
02
Press CapsLock.
YapClean listens while you speak naturally, including pauses and corrections.
03
Speak like a person.
No need to compose polished sentences in your head before talking.
04
Get clean text.
The result is cleaned for clarity and inserted at the cursor in supported fields.
YapClean is in beta. Some Windows fields behave differently, and compatibility will improve as testing expands.
Try the beta when your workflow is ready for cleaner dictation.
Tell us how you plan to use YapClean, which AI provider you prefer, and whether you want Free BYOK, Founder Lifetime BYOK, or the Managed plans waitlist.