Practise Interview Answers and Get a Transcript Without Uploading

Why this works

Hearing yourself is uncomfortable and useful. A local transcript lets you count filler words, measure answer length, and check whether your examples have a clear result.

Step by step

1. Pick five likely questionsUse the questions for your role and write them on cards.
2. Record each answerAnswer aloud for about two minutes per question, with your phone on a table, and save each file.
3. Transcribe locallyConvert your recording to 16 kHz mono WAV: ffmpeg -i input.m4a -ar 16000 -ac 1 -c:pcm_s16le audio.wav. Then run whisper.cpp on each file. Build 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.
4. Score the transcriptCount filler words (um, like, you know), note the number of seconds before your first point, and check that each answer has a result.
5. Re-record the weakest answerImprove one answer per session rather than rewriting all five.

Copy-paste prompt

Review this practice interview answer. Report: word count, filler words with counts, whether there is a clear situation, action, and result (quote where each is), and one suggestion to shorten it. Do not rewrite the answer for me.

ANSWER TRANSCRIPT:
<paste>

Check your result

• Each answer includes an outcome you can verify
• Filler-word counts were checked by ear on one recording
• Your examples are true and can be discussed in depth

Pitfalls to avoid

• Do not memorise scripted answers word for word; aim for structure and confidence.
• A transcript cannot judge tone or body language. Also record a short video once in a while.