Escribolt

Escribolt is designed as a native desktop application optimized for macOS. For maximum fluid performance and low latency, review the following prerequisites:

Operating System

Escribolt is fully compatible with the following macOS releases:

  • macOS 15 (Sequoia)
  • macOS 14 (Sonoma)
  • macOS 13 (Ventura)
  • macOS 12 (Monterey)

Hardware Architecture

Escribolt utilizes hardware acceleration for local models. Requirements vary depending on your processor:

  • Apple Silicon (M1, M2, M3, M4 series): Highly recommended. Escribolt's Whisper.cpp transcription pipelines are optimized to compile and execute directly on Apple's Neural Engine (ANE) and Unified Memory.
  • Intel Processors (Core i5, i7, i9, Xeon): Fully supported. However, local model inference will utilize CPU cores, leading to higher latency and memory consumption compared to Apple Silicon.

Disk Space Recommendations

  • Core Client: ~150MB of disk space for the compiled desktop application.
  • Local Models Bootstrap: ~2.0GB of free space is required if you plan to activate Local Mode. This space hosts quantized weights for local Speech-to-Text (Whisper Base) and reasoning models (Qwen-2.5 / Gemma).