Setting up Whisper locally for unlimited free dictation

OpenAI's Whisper model is state-of-the-art for speech recognition. While the API is great, running it locally is even better. Thanks to the work of Georgi Gerganov on whisper.cpp, we can run these models on consumer hardware with incredible performance.

Interactive Setup Guide

We've built this tool to generate the exact commands you need based on your machine and preferences. Select your OS and desired model quality below:

# 1. Clone the repository
git clone https://github.com/ggerganov/whisper.cpp.git
cd whisper.cpp

# 2. Build the project
# Optimized for Apple Silicon (Metal)
make

# 3. Download Model
bash ./models/download-ggml-model.sh base.en

# 4. Run Transcription
./main -m models/ggml-base.en.bin -f samples/jfk.wav

The generator above handles the cloning, building, and model downloading steps. Once you've run those commands, you'll have a working main executable in your directory.

Real-time Dictation

For real-time usage, tools like Andak wrapper these libraries into a user-friendly interface that injects text directly into your active application, effectively replacing the system dictation with a smarter AI brain.