Projects / Voice and audio

Turn voice memos into text

Drop in a recording of a meeting or voice memo and get a written transcript, all on your own computer.

At a glance

You'll get
A one-line command that turns any audio file into a text file next to it.
Time
About 30 minutes
Difficulty
Easy
Computers
Regular laptop or desktop (16 GB), Gaming PC (NVIDIA graphics card), Newer Mac (32 GB or more)
Cost
Free
Under the hood
Whisper speech-to-text via whisper.cpp or openai-whisper. Models: OpenAI Whisper (size picked per computer)

Know the limits medium risk

  • It mishears names, numbers, jargon, and strong accents, and can drop or invent words in noisy audio.
  • Check the transcript against the recording before you share it or rely on it, especially for quotes, figures, or anything legal or medical.

More in Using local AI safely.

Check the results before you rely on them.

The easy way: have an AI assistant do itRecommended if you're new to this

Open an AI coding assistant on the computer you want to set up (Claude Code, Codex, and OpenCode all work), paste the instructions below, and follow along. It will check your computer, install what's needed, and confirm everything works.

What's an AI coding assistant?

It's a version of ChatGPT-style AI that can run commands on your computer for you, with your permission. You talk to it in plain English, and it does the typing.

Before you paste: this lets an AI assistant run commands on your computer. Read what it plans to do before you approve each step, and say no to anything you don't understand.

Install local speech-to-text on this machine using Whisper (prefer whisper.cpp with CUDA or Metal when available; otherwise a CPU build or openai-whisper).

Goals:
1. Detect OS and accelerators (NVIDIA CUDA, Apple Metal, or CPU-only).
2. Install ffmpeg if missing (use the OS package manager; ask before sudo).
3. Install whisper.cpp or openai-whisper and download a model that fits available memory (`base` is a safe default on unknown hardware).
4. Transcribe a short sample file I provide (or record a few seconds if I agree).
5. Write the transcript to a text file next to the audio and show me the path.

Constraints:
- Keep audio and transcripts on this machine only.
- Do not publish speed claims beyond what you measure on this run.
- Document the exact model filename and commands in a short local README.

When done, give me the one-liner to transcribe future files.

Do it yourselfFor hands-on folks, or if you want to learn how it works

Before you start

Have a short recording ready, a minute or two of clear speech. Phone voice memos work fine.

What does this mean?

Whisper is a free speech-recognition model released by OpenAI. It comes in sizes from "tiny" to "large". Bigger is more accurate but slower. On a regular laptop start with "base"; with a graphics card (GPU) or a newer Mac you can go bigger.

Do it yourself

  1. Install ffmpeg, a free tool that converts audio formats.
  2. Install whisper.cpp (or the openai-whisper Python package).
  3. Download a model file, for example ggml-base.bin.
  4. Transcribe your recording: bash whisper-cli -m models/ggml-base.bin -f memo.wav -otxt
  5. Open the .txt file that appears next to your recording.

How to tell it worked

  • The text matches what was said, give or take a few words.
  • It runs with Wi-Fi off.

Test results

Results pending. We'll publish speed and quality on each of our test computers once we've run it. Until then, no numbers.