Licenses & Third-Party Components
The App uses the following open-source software and models. The full text of each license is also bundled in the app (Whisper Local.app/Contents/Resources).
Software
- Argmax OSS Swift (WhisperKit / SpeakerKit / ArgmaxCore) — MIT License — github.com/argmaxinc/argmax-oss-swift
- swift-transformers (the origin of some code in Argmax OSS Swift) — Apache License 2.0 — github.com/huggingface/swift-transformers (notice in the bundled Argmax-NOTICES.txt)
- swift-argument-parser (build dependency) — Apache License 2.0 with Runtime Library Exception — github.com/apple/swift-argument-parser
Models (not bundled with the app; downloaded from the distributor when the user chooses to)
- OpenAI Whisper large-v3-turbo (standard and Compact versions) / base.en (Core ML conversions by Argmax) — MIT License — huggingface.co/argmaxinc/whisperkit-coreml / github.com/openai/whisper
- SpeakerKit Pyannote Core ML — CC-BY-4.0 (as stated by the distributor) — huggingface.co/argmaxinc/speakerkit-coreml
Follow each distributor’s stated terms for using the models. The App does not modify the model files.
macOS
SwiftUI, AppKit, AVFoundation, Core ML, Speech, NaturalLanguage, Network, Security, CryptoKit, ServiceManagement and CoreAudio are standard macOS features. Python and ffmpeg are neither bundled nor used.