Skip to content

AI Agent Prompt-Based Editing

Edit media using plain English instructions. An AI agent interprets your prompt and creates a precise edit plan.

Usage

praisonai-editor edit interview.mp3 \
  --prompt "Remove the intro, any weather discussion, and keep only the technical part"

How it works

sequenceDiagram
    participant You
    participant CLI
    participant Whisper
    participant Agent
    participant FFmpeg

    You->>CLI: edit file.mp3 --prompt "..."
    CLI->>Whisper: transcribe (with cache)
    Whisper-->>CLI: transcript + timestamps
    CLI->>Agent: transcript + prompt → create edit plan
    Agent-->>CLI: JSON segments (keep/remove)
    CLI->>FFmpeg: render keep segments
    FFmpeg-->>You: edited file ✓

Prompt examples

praisonai-editor edit podcast.mp3 \
  --prompt "Remove the first 5 minutes of introduction"

praisonai-editor edit interview.mp3 \
  --prompt "Remove any discussion about weather or personal anecdotes, keep technical content only"

praisonai-editor edit lecture.mp3 \
  --prompt "Keep only the question and answer section at the end"

Fallback

If the agent cannot parse its own output, it falls back to the heuristic podcast preset automatically.

Python API

from praisonai_editor.agent_pipeline import prompt_edit

result = prompt_edit(
    "interview.mp3",
    "Remove the intro and keep only technical discussion",
    output_path="interview_edited.mp3",
    verbose=True,
)