GPT-4o
⚙️ Technical
Intermediate
Offline Transcription Pipeline Setup
Set up a fully offline audio and video transcription system using Whisper and local post-processing.
The Prompt
# Offline Transcription Pipeline Setup You are a local AI engineer. Design and set up an offline transcription pipeline for the described audio workload. ## Workload Details - Audio source: [MEETING RECORDINGS / PODCASTS / INTERVIEWS / CALL CENTER AUDIO / OTHER] - Language(s): [PRIMARY LANGUAGE, plus any secondary] - Average file duration: [MINUTES PER FILE] - Files per week: [VOLUME] - Output needed: [TRANSCRIPT ONLY / TRANSCRIPT + SUMMARY / TRANSCRIPT + SPEAKER LABELS] - Hardware: [CPU AND GPU SPECS] ## Pipeline Design ### Model Selection Choose the right Whisper variant (tiny / base / small / medium / large-v3 / turbo) for this hardware and accuracy requirement. Explain the trade-off. ### Installation Steps Exact commands for setting up faster-whisper or whisper.cpp on [OPERATING SYSTEM]. ### Batch Processing Script A Python or shell script skeleton that processes all files in a folder, outputs .txt transcripts, and logs errors. ### Post-Processing If [OUTPUT NEEDED] includes summaries or speaker labels, describe the additional local model step. ### Expected Performance Estimate transcription speed (real-time factor) for [AVERAGE FILE DURATION] on the specified hardware.
📝 Fill in the blanks
Replace these placeholders with your own content:
[MEETING RECORDINGS / PODCASTS / INTERVIEWS / CALL CENTER AUDIO / OTHER]
[PRIMARY LANGUAGE, plus any secondary]
[MINUTES PER FILE]
[VOLUME]
[TRANSCRIPT ONLY / TRANSCRIPT + SUMMARY / TRANSCRIPT + SPEAKER LABELS]
[CPU AND GPU SPECS]
[OPERATING SYSTEM]
[OUTPUT NEEDED]
[AVERAGE FILE DURATION]
How to use this prompt
1
Copy the prompt
Click "Copy Prompt" above to copy the full prompt text to your clipboard.
2
Replace the placeholders
Swap out anything in [BRACKETS] with your specific details.
3
Paste into GPT-4o
Open your preferred AI assistant and paste the prompt to get started.