Prompt
Initial Prompt
The original instruction used to turn the source footage into a tightly paced rough cut.
Install HyperFrames from:
https://github.com/heygen-com/hyperframes
Follow the repository installation instructions.
Use the Groq API for transcription:
API KEY = <YOUR_GROQ_API_KEY>
Apply rough cut to the video in the current working folder.
Make the pacing feel like a tightly edited YouTube video:
- Start every retained clip 20–30 ms before the speaker’s first audible phoneme.
- End every clip 50–70 ms after the final audible phoneme.
- Remove breaths, hesitation, idle frames, and dead air before or after the dialogue.
- Keep the total audible gap between consecutive lines within 100–180 ms.
- Do not cut directly on the waveform onset or ending; preserve the complete consonant and vowel tails.
- Use 10–20 ms audio crossfades where needed to prevent clicks or abrupt room-tone changes.
- Review every boundary by listening and inspecting the waveform. Do not rely only on transcript timestamps.
The result should feel immediate and energetic: the speaker begins talking as soon as each shot appears, without clipped words or noticeable silence.
Add your own Groq API key after copying. Credentials are intentionally excluded from this public resource.