Subtitles without the cloud: turning speech into text locally with whisper.cpp
A local transcript is not yet a finished subtitle track. A short recording of your own and a WebVTT example show how wording, translation and timing fit together.
A local transcript is not yet a finished subtitle track. A short recording of your own and a WebVTT example show how wording, translation and timing fit together.
When importing a PDF, a lost column can change the meaning. A reference document you create yourself shows what a structured export has to preserve.
Quantisation changes how the numbers in model weights are represented. A tiny rounding experiment explains the principle before you compare large model files.
What makes an AI usable offline is prepared files and a clearly chosen working route. A course sheet of your own provides a manageable test for chat, a document question and a local server.
A local interface does not yet tell you where all data is processed. With local-only mode, three clear stations and a text sample of your own, Ollama's working route becomes verifiable.
The most useful starting question is not which ai is more impressive.
Practical steps on the KENJTEC platform.
KENJTEC · Feature note
Open your WordPress files in the browser and compare earlier file versions before continuing your work.
Read the guideKENJTEC · Feature note
In DomainStudio, you edit your domain’s DNS records with explanations and a check before saving.
Read the guideKENJTEC · Feature note
In SocialStudio, you create versions for your channels, edit the texts and then publish them yourself.
Read the guide
Hamburg is our starting point.
Practical AI from Hamburg, explained for everyone.