Skip to content

Argmax WhisperKit

On-device speech recognition for iOS and macOS apps

Visit official site

What Argmax WhisperKit is

The toolkit is described as a Whisper-based inference toolkit from Argmax that enables speech recognition and transcription inside iOS and macOS applications. The project is open source and stated to help developers and companies run commercial-scale inference workloads on user devices.

What you can do with it

  • Add Whisper inference to an app through the Swift package
  • Try the provided test applications on iOS and macOS
  • Optimize and evaluate Whisper on Mac with the Python tools
  • Run speech recognition and transcription inside your own app

Project notes

  • The project is described as open source
  • The stated aim is to collect developer feedback
  • A stable release candidate is planned within weeks
  • The goal is to speed up production use of on-device inference

Example uses

  • Transcribe speech in real time within an iOS application
  • Integrate the toolkit into a macOS app for speech recognition
  • Use the Python tools to tune Whisper performance

Pros & cons

✓ What we like

  • The project is open source
  • A release candidate is planned within weeks according to the material

! What to watch out for

  • The stated goal is to gather developer feedback before a stable release candidate
  • The toolkit targets iOS and macOS applications

FAQ

What platforms are named?

The material names iOS and macOS applications, with Mac used for the Python tools.

What is included in the toolkit?

A Swift package, iOS and macOS test apps, and Python tools for optimizing and evaluating Whisper on Mac.

Last reviewed: 2026-11-15

More AI audio tools tools

View all →

How we review