Repolode
Submit
argmaxinc

argmaxinc/argmax-oss-swift

Swift SDK for running speech-to-text, text-to-speech and speaker diarization fully on-device on Apple silicon.

Open-source alternative to:Otter.ai

View on GitHub
Stars
6,411
Forks
622
Open issues
143
Last commit
10 days ago

What is it?

Argmax Open-Source SDK for Swift bundles three on-device speech frameworks in one Swift package: WhisperKit for speech-to-text with OpenAI Whisper models, TTSKit for text-to-speech with Qwen3-TTS models, and SpeakerKit for speaker diarization with Pyannote models. All of them run on Core ML on Apple silicon, so audio is processed locally without a cloud server.

WhisperKit can transcribe local audio files or a microphone stream, and it downloads Whisper variants from Hugging Face on demand. The Argmax CLI includes a local server that implements the OpenAI Audio transcription and translation endpoints, so existing OpenAI SDK clients can point at it. SpeakerKit can merge speaker labels with WhisperKit transcripts and export RTTM output.

The package targets macOS 14 and later for the Swift package as a whole, with TTSKit requiring macOS 15 or iOS 18. It is distributed under the MIT license, and a commercial Argmax Pro SDK with additional models and Android support is offered separately.

Key features

  • 01Speech-to-text with Whisper models via WhisperKit, including streaming from a microphone
  • 02Text-to-speech with Qwen3-TTS models and real-time streaming playback in TTSKit
  • 03Speaker diarization with Pyannote models and RTTM export in SpeakerKit
  • 04Local OpenAI-compatible Audio API server for transcription and translation
  • 05Swift CLI (argmax-cli) and Homebrew install for whisperkit-cli

Use cases

  • Adding offline transcription to an iOS or macOS app without sending audio to a server
  • Generating speech on-device in a Swift app with streaming playback
  • Labeling who spoke when in meeting or interview recordings
  • Pointing existing OpenAI SDK code at a local transcription server on a Mac

Quick start

  1. 1

    Add the Swift package dependency in Xcode or Package.swift (Swift Package Manager)

  2. 2

    Install the CLI via Homebrew

    brew install whisperkit-cli

Things to know

  • Requires macOS 14.0 or later and Xcode 16.0 or later.
  • Targets Apple silicon on-device inference only, not Android or other platforms.

Topics

#speech-recognition#text-to-speech#speaker-diarization#on-device#core-ml#swift#whisper#apple-silicon#inference#ios

FAQ

What is argmax-oss-swift?
Swift SDK for running speech-to-text, text-to-speech and speaker diarization fully on-device on Apple silicon. Argmax Open-Source SDK for Swift bundles three on-device speech frameworks in one Swift package: WhisperKit for speech-to-text with OpenAI Whisper models, TTSKit for text-to-speech with Qwen3-TTS models, and SpeakerKit for speaker diarization with Pyannote models. All of them run on Core ML on Apple silicon, so audio is processed locally without a cloud server.
Is argmax-oss-swift free to use? What license does it have?
argmax-oss-swift is open source under the MIT license. Check the license terms before using it in commercial products.
Is argmax-oss-swift still maintained?
Actively maintained. The last commit to argmax-oss-swift was on Oct 11, 2026.
What can I use argmax-oss-swift for?
Adding offline transcription to an iOS or macOS app without sending audio to a server Generating speech on-device in a Swift app with streaming playback Labeling who spoke when in meeting or interview recordings Pointing existing OpenAI SDK code at a local transcription server on a Mac
How do I install argmax-oss-swift?
Add the Swift package dependency in Xcode or Package.swift (Swift Package Manager) Install the CLI via Homebrew: brew install whisperkit-cli
Where can I get argmax-oss-swift?
The source code and installation instructions are on GitHub: https://github.com/argmaxinc/argmax-oss-swift

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects