System Requirements

What you need

• macOS with Apple Silicon, M1 or newer (M1 through M4). Intel Macs are not supported.
• GPU acceleration, many times faster than the CPU.
• ±2–4 GB free space for language models, downloaded once, then locally cached.
• Transcription runs fully locally, internet needed for the first download and periodic licence check (with offline grace period).
• FFmpeg included, no separate installation needed.

Speech recognition & speed

Our pipeline runs on advanced, locally accelerated speech recognition. The model is automatically tuned to the best balance of speed and quality, downloaded once, then fully local.

Chip	10 min video ≈
M1 / M1 Air	±5–7 min
M2 / Pro / Max	±2–4 min
M3 / Pro / Max	±1.5–3 min
M4 / Pro / Max	±1–2 min

Estimates, actual speed depends on audio quality and model choice. ChatGPT correction (Pro) uses your own OpenAI key and charges per token.