Guides and engineering notes from building Tota — practical transcription how-tos, and what it takes to run Whisper and a language model entirely on a Mac. Written by the developer, verified against the shipping app.
How to Transcribe an Interview on a Mac — Offline, With Speaker Names
Turn an interview recording into a speaker-labelled transcript entirely on your Mac — no upload, no per-minute pricing. Step-by-step with export formats, accuracy tips, and why offline matters for source protection.
Running Qwen 3 on Apple Silicon with MLX: What a 2.3 GB LLM Can Do
Why Tota ships a 4-bit Qwen 3 4B on MLX instead of calling a cloud API: model choice, quantization, unified memory, latency, and the guardrails that keep a small local LLM honest.
How to Transcribe Apple Voice Memos on a Mac
Turn any Apple Voice Memo into a text transcript on your Mac — offline, with automatic speaker labels, and no upload. Step-by-step, plus when Apple's built-in transcript is enough.
Whisper Model Sizes Explained: Which One Should Your Mac Run?
Tiny, Base, Small, Medium, Large v3, Large v3 Turbo — what actually changes between Whisper model sizes, what each needs from your Mac, and which one to pick for dictation vs file transcription.
