Common Questions

A macOS app that transcribes audio and video files and generates AI summaries. Everything runs locally on your Mac — no cloud, no account needed.

A Mac with Apple Silicon (M1, M2, M3, or M4) running macOS 14 or later. We recommend 16 GB of RAM for best performance (8 GB minimum). You'll also need about 5 GB of free disk space for the AI models.

On first launch, Murmur downloads two AI models — a speech recognition model (~150 MB) and a summarization model (~4 GB). This only happens once. After that, everything runs instantly from your local storage.

Only on first launch to download the AI models. After that, transcription and summarization work completely offline. The only other network requests: license checks and software updates.

MP3, M4A, and WAV for audio. MP4 and MOV for video. Video files are automatically converted to audio. The original file is not modified.

All transcripts and recordings are stored locally with AES-256-GCM encryption. No data leaves your Mac. No telemetry, no tracking, no cookies.

No. All audio and video files, transcripts, and summaries stay on your Mac. No uploads, no cloud processing, no third-party access.

Yes. All data stays local on your device. No personal data is collected, no user accounts are created, and no telemetry is sent. There is no data processor in the chain.

Only three: license verification via Lemon Squeezy (transmits only the license key, no content), software update checks, and the one-time AI model download on first launch.

Yes. That's exactly what Murmur is built for. Legal consultations, research interviews, medical dictations, personnel decisions — everything stays local and encrypted on your Mac.

Yes. Murmur automatically identifies and labels different speakers in the transcript — including in meetings with multiple participants, interviews, and focus groups.

Yes. Murmur captures your Mac's system audio — regardless of which video conferencing tool you use. No browser extension, no meeting bot, no screen sharing needed.

A local language model (Qwen 3.5, 9 billion parameters) generates summaries directly on your Mac. The summary adapts to the content type — meeting minutes, lecture notes, interview highlights, or podcast overview.

Yes. You can regenerate a summary with a different focus at any time, without re-processing the recording.

Transcripts and summaries can be exported as TXT, DOCX, or PDF. Or copy to clipboard in one click for Notion, Confluence, email, or other tools.

Two models, both running locally: Parakeet ASR (NVIDIA NeMo-based) for speech recognition and Qwen 3.5 (9B parameters, 4-bit quantized) for summarization. Both use your Apple Silicon chip's Neural Engine and GPU.

The AI models use the Neural Engine and Metal GPU acceleration available only on Apple Silicon (M1, M2, M3, M4). Intel Macs are not supported.

25+ languages with automatic detection. You don't need to select the language manually — Murmur detects it automatically.

Depends on the recording length and your Mac. On an M1 with 16 GB RAM, one hour of audio typically takes 5-15 minutes.

One-time purchase for €29 — no subscriptions, no hidden fees, no per-minute charges, no usage limits.

7 days with full functionality. No credit card required. After the trial, you can still open and export existing transcriptions.

Up to three. You can deactivate the license in Settings and transfer it to a different Mac.

Yes. 30-day money-back guarantee after purchase, no questions asked.