MinuteWave

Local-first AI meeting copilot for macOS.

Capture meeting audio, generate high-quality transcripts with speaker labels, create structured summaries, and chat with transcript-grounded citations.

Download latest DMG · Permissions setup guide

Why MinuteWave

MinuteWave is built for people who want native macOS meeting notes without a browser-first workflow:

• Local transcription path with FluidAudio (Parakeet v3 + offline diarization).
• Cloud transcription options with Azure OpenAI or OpenAI.
• Transcript-aware AI chat and summary generation.
• Session persistence, export, and optional database encryption.
• Native SwiftUI UI with onboarding, settings, and update checks.

Feature Overview

Area	What you get
Audio capture	Microphone only or Microphone + system audio capture modes
Transcription	Local FluidAudio, Azure OpenAI Whisper, or OpenAI Whisper
Speaker labels	Offline diarization in local FluidAudio mode
Summary	Structured markdown summary generation from transcript content
Transcript chat	Q&A with lexical retrieval and timestamp citations
Export	Markdown, TXT, and JSON export per session
Storage	Local SQLite session store + Keychain-backed secrets
Security	Optional SQLCipher encryption mode with migration support
UX	EN/NL app localization, onboarding wizard, update checker

Provider Matrix

Capability	Local FluidAudio	Azure OpenAI	OpenAI	LM Studio
Transcription	Yes	Yes	Yes	No
Summary	No	Yes	Yes	Yes
Transcript chat	No	Yes	Yes	Yes
API key required	No	Yes	Yes	Optional

Architecture

flowchart LR
  A["Microphone"] --> B["HybridAudioCaptureEngine"]
  C["System audio (optional)"] --> B
  B --> D["Transcription provider"]
  D --> E["SQLite session repository"]
  E --> F["Summary provider"]
  E --> G["Transcript chat + citations"]
  E --> H["Export service (MD/TXT/JSON)"]

  subgraph Providers
    D1["Local FluidAudio"]
    D2["Azure OpenAI"]
    D3["OpenAI"]
  end

  subgraph Summary/Chat
    S1["Azure OpenAI"]
    S2["OpenAI"]
    S3["LM Studio"]
  end

  D --- D1
  D --- D2
  D --- D3
  F --- S1
  F --- S2
  F --- S3
  G --- S1
  G --- S2
  G --- S3

Loading

System Requirements

• macOS 14+
• Apple Silicon Mac
• 16 GB RAM minimum (enforced by onboarding checks)
• Internet connection for cloud providers and first local model download

Installation (DMG)

1. Download from latest release.
2. Open MinuteWave-macOS.dmg.
3. Drag MinuteWave.app to Applications.
4. Start with right-click -> Open on first run.

Important

If Gatekeeper blocks launch, use one of these options:

1. System Settings -> Privacy & Security -> Open Anyway
2. xattr -dr com.apple.quarantine "/Applications/MinuteWave.app"

Quick Start

1. Complete onboarding permissions (Microphone, and optionally Screen Recording for system audio capture).
2. Choose your transcription provider (Local (FluidAudio) for local-first workflow, or Azure/OpenAI for cloud transcription).
3. (Optional) Configure summary/chat provider (Azure, OpenAI, or LM Studio).
4. Create a session, start recording, stop recording.
5. Review transcript, generate summary, ask follow-up questions, export results.

Privacy & Security

• Session data is stored locally in ~/Library/Application Support/MinuteWave.
• API keys are stored in macOS Keychain.
• Database encryption can be enabled in settings when SQLCipher runtime is available.
• Encryption migrations (plaintext <-> SQLCipher) are built in.
• Local FluidAudio mode keeps inference on-device after model preparation.

Build From Source

Prerequisites

• Xcode (latest stable)
• Swift 6 toolchain
• Homebrew packages:

brew install sqlcipher create-dmg

Build and run

swift build
swift test
swift run MinuteWave

For reliable macOS permission prompts, run as a real app bundle:

./scripts/build_dev_app_bundle.sh debug
open ".build/AppBundle/MinuteWave.app"

Build DMG

./scripts/build_dmg.sh release

Optional signing:

security find-identity -v -p codesigning
SIGNING_IDENTITY="Apple Development: Your Name (TEAMID)" ./scripts/build_dev_app_bundle.sh release

Release and CI

• Release workflow: .github/workflows/release.yml
• Tag format: vMAJOR.MINOR.PATCH
• Release artifacts: MinuteWave-macOS.dmg, MinuteWave-macOS.dmg.sha256

Troubleshooting

• Permission issues: see docs/XcodePermissionsSetup.md
• Reset TCC permissions:

./scripts/reset_tcc_permissions.sh

• LM Studio model not detected: refresh status in Settings -> AI tab and ensure at least one model is loaded.

Star History

License

MIT License. See LICENSE.