Loading page...Загружаем страницу...正在加载页面...
AtlasRepo

I didn't want to pay a monthly subscription for dictation, so I built one that runs on my GPU

Description and use

Wisperno is an offline dictation app for Windows PCs with NVIDIA GPUs, avoiding subscriptions and network traffic. It uses Whisper large‑v3‑turbo and a quantized language model, delivering 0.82‑1.48 s latency for 100 words and 13.5× real‑time speed in live transcription mode.

Problem

Need offline, subscription‑free dictation on a PC

Outcome

Runs fully offline on NVIDIA GPU, 0.82‑1.48 s latency for 100 words, 97.6‑98.8 % fidelity, 13.5× faster than real time in live mode

Workflow

  • Hold hotkey
  • Speak
  • Release
  • Text pasted into focused app
  • Optional polish via local model

Evidence

It runs fully offline. Whisper large‑v3‑turbo transcribes on the GPU, and a small quantized language model cleans up the text.

Source

A long pause (1.3 seconds or more) starts a new paragraph. I first tried segment‑level timestamps from Whisper to detect the pause, and they could not see it at all. Word‑level timestamps can.

Source

The smaller presets are faster and use less memory. The bigger ones keep a little more fidelity: 97.6% for the two smallest, 98.8% for the two 3B models, measured on raw model output.

Source

Live Transcription is a separate mode for meetings and lectures. My first design re‑decoded only a sliding window of the latest audio. The test suite showed it silently dropped real words. The current version re‑decodes the whole unconfirmed audio buffer every 0.6 seconds. It runs about 13.5 times faster than real time (RTF 0.074), with no second model loaded.

Source

  • Record date: 2026-10-21T14:41:05.380Z