I built Yapper because I love the accuracy of Whisper, but I was frustrated that the best native macOS wrappers required monthly subscriptions.
The premise is simple: If the model runs locally on my hardware (Apple Silicon), I shouldn't have to pay rent on the software.
Yapper is a native macOS app that runs entirely offline. You press a hotkey, speak, and it types into any text field (VS Code, Slack, Notes, etc).
How it works: It’s built on WhisperKit and optimized for the Apple Neural Engine. Because it runs locally, your voice data never leaves your machine.
Features:
- 100% Offline: Zero latency issues, total privacy.
- No Subscription: A simple one-time purchase.
- Optional AI Polish: While transcription is offline, I added a "Bring Your Own Key" feature (supports Claude, OpenAI, Gemini). If enabled, it sends the text (not audio) to the LLM of your choice to fix grammar or restructure rambling before pasting it.
I’m launching this with a lifetime license for $24 (usually $49) to get early feedback. I tried to build the exact tool I wanted to use: fast, private, and paid for once.
I’d love to hear your thoughts on the implementation or features you’d like to see next!
0 comments
—
Show HN: Yapper – Offline macOS dictation. One-time purchase, no sub · HackerTrans
I built Yapper because I love the accuracy of Whisper, but I was frustrated that the best native macOS wrappers required monthly subscriptions.
The premise is simple: If the model runs locally on my hardware (Apple Silicon), I shouldn't have to pay rent on the software.
Yapper is a native macOS app that runs entirely offline. You press a hotkey, speak, and it types into any text field (VS Code, Slack, Notes, etc).
How it works: It’s built on WhisperKit and optimized for the Apple Neural Engine. Because it runs locally, your voice data never leaves your machine.
Features:
- 100% Offline: Zero latency issues, total privacy.
- No Subscription: A simple one-time purchase.
- Optional AI Polish: While transcription is offline, I added a "Bring Your Own Key" feature (supports Claude, OpenAI, Gemini). If enabled, it sends the text (not audio) to the LLM of your choice to fix grammar or restructure rambling before pasting it.
I’m launching this with a lifetime license for $24 (usually $49) to get early feedback. I tried to build the exact tool I wanted to use: fast, private, and paid for once.
I’d love to hear your thoughts on the implementation or features you’d like to see next!