Verified reference stack

Consultant — Private Meeting Transcription — RTX 4060

whisper.cpp documents large-family memory around ~3.9 GB and provides large-v3-turbo model artifacts; 8 GB VRAM is a comfortable reference profile for local single-stream transcription.

Application

whisper.cpp

Runtime

whisper.cpp

Model

Whisper large-v3-turbo

Hardware

RTX 4060 8 GB / 32 GB RAM

GOOD

Privacy

Fully local

License / terms

MIT for OpenAI Whisper code and model weights

Evidence

https://github.com/openai/whisper
https://github.com/ggml-org/whisper.cpp/blob/master/README.md
https://github.com/ggml-org/whisper.cpp/blob/master/models/README.md

Freshness rule: Material changes set this stack to REVALIDATION_REQUIRED.