whisperx

Prerequisites

Install and load these skills before generating (skip if already in context via @pruna):

Or install the full suite once: npx skills add PrunaAI/pruna-skills@pruna -y

Follow each skill's Before generating / craft sections — do not restate guide content here.

Agent habit

In the first reply, name `whisperx` in backticks, confirm REPLICATE_API_TOKEN (or stop with signup links from pruna-api), then ask for required inputs. Open intake → generation-diversity clarification intake before the first POST. Redirect when When NOT to use fits better.

When NOT to use

Use a different skill instead:

Environment

export REPLICATE_API_TOKEN=r8_...

HTTP / pipeline

Prefer the repo helpers (they upload local audio and write JSON + SRT):

# Agent: follow phase table in this SKILL.md \
  --song output/my-mv/song.mp3 \
  --out output/my-mv/whisperx_transcript.json \
  --initial-prompt "First few lyric lines help recognition"

# Agent: follow phase table in this SKILL.md \
  --cuts output/my-mv/cut_manifest.json \
  --transcript output/my-mv/whisperx_transcript.json \
  --song output/my-mv/song.mp3

In a music-video workflow, run this skill during the align phase — see music-video.

Before generating

1. Complete Prerequisites guide reading order.
2. Confirm audio_file (HTTPS URL — helpers upload local files), language, and align_output: true for word-level cut alignment. When listing inputs, name REPLICATE_API_TOKEN (Replicate — not PRUNA_API_KEY).
3. Model notes: set initial_prompt to the first lyric lines for better rap/sung recognition. Optional diarization for multi-voice battles.

Required input

• audio_file (HTTPS URL)

Common optional fields

• language — ISO code, e.g. en
• align_output — true for word-level timestamps (required for cut alignment)
• initial_prompt — first lyric lines
• diarization — speaker labels

Typical next steps

Common follow-ons after this skill: