Turn Lecture Recordings into Captioned Study Notes Offline

Why this works

Many courses record lectures, and re-listening is slow. A transcript you can search turns an hour of audio into minutes of review.

Step by step

1. Set up whisper.cppBuild whisper.cpp from the repository (follow its README for CMake steps), then download a model with the bundled script, for example ./models/download-ggml-model.sh base.en. Models download once; transcription runs offline.
2. Convert and transcribeConvert your recording to 16 kHz mono WAV: ffmpeg -i input.m4a -ar 16000 -ac 1 -c:pcm_s16le audio.wav. Then run the transcription binary and keep the timestamps in the output.
3. Generate notes from the transcriptUse the prompt below in a local model, one lecture at a time.
4. Link notes back to timestampsKeep the [00:12:31] markers so you can jump to the exact point in the audio when you revise.
5. Review with your own wordsRewrite three key ideas per lecture without looking at the notes. That is the step that builds memory.

Copy-paste prompt

Turn this lecture transcript into study notes: a title, five key ideas as bullets with the timestamp they start at, three terms with plain-English definitions taken from the lecture, and two questions a student should be able to answer afterwards. Do not add facts that are not in the transcript.

TRANSCRIPT:
<paste>

Check your result

• Timestamps match the audio
• Definitions come from the lecture, not from memory of the subject
• You can explain each key idea without the notes

Pitfalls to avoid

• Check recording rules with your instructor or institution before recording a class.
• Technical terms are often misheard; correct them against the slides.