Open Source · v0.1.0

From interview
to timeline

AI-assisted editorial intelligence for documentary and brand filmmakers. Transcribe, analyze emotional delivery, build narrative arcs, and export to DaVinci Resolve — all running locally on your machine.

View on GitHub → See the Pipeline
Scroll
The Pipeline
Six stages from raw footage to rough cut
Each stage runs independently. Re-run any step without reprocessing what came before. All intermediate data is JSON — inspectable, editable, yours.
🎬
Extract
Audio + metadata from any video format
📝
Transcribe
Word-level timestamps via Whisper
🎭
Analyze
Energy, pitch, pace, pauses — how it was said
🧠
Themes
LLM extracts themes across all interviews
📐
Arc
Narrative structure with editorial reasoning
🎞️
Export
EDL / FCPXML straight into Resolve
Core Differentiator
It's not just what they said. It's how they said it.
Plotline analyzes the audio alongside the transcript — energy, pitch variation, speech rate, pauses. The moments that give you chills? Now they have a score.
Interview 03 — Delivery Analysis
Energy
Speech Rate
Pause
00:00:00 00:12:30 00:25:00 00:37:30 00:50:00
00:18:33.200
"That was the last time I saw her at the river." — Voice drops, 2.8s pause before. Delivery: quiet, deliberate, weighted.
00:31:07.440
"Now I bring my own grandchildren every morning." — Energy rises, pitch lifts. Delivery: warm, purposeful, hope.
00:42:15.880
"The water remembers everything. That's what she taught us." — Measured pace, rising conviction. Full circle moment.
Features
Built for how editors actually work
🔒
Privacy
Local-first. Always.
Transcription and audio analysis run entirely on your machine. Your audio never leaves your SSD. Cloud LLMs are optional — and only for projects where you choose to allow it.
🔄
Cross-Interview
Find the thread across voices
When five people talk about the same thing, Plotline finds the connections. Shared themes, complementary perspectives, contradictions — surfaced automatically across all interviews in your project.
📋
Creative Brief
Brief in, selects out
Feed in your key messages, audience, and tone direction. Plotline evaluates every segment against your brief and shows you a coverage matrix — what's covered, what's weak, what's missing.
⚖️
Best Take
Who said it best?
When multiple speakers address the same message, Plotline ranks their versions by delivery quality, content alignment, and conciseness. Side-by-side comparison with audio playback. You pick the winner.
✏️
Your Prompts
The editorial intelligence is yours
Theme extraction, narrative structure, selection criteria — all controlled through editable prompt templates. Change how Plotline thinks about story without touching code. Your editorial judgment, amplified.
✅
Review Gate
Nothing hits the timeline without you
Approve, reject, reorder, annotate. Listen to every segment with audio playback. Keyboard-driven review: arrow down, spacebar, A to approve, X to reject. 60 segments in 10 minutes.
Project Profiles
One tool, three modes
Each profile configures the entire pipeline — delivery weights, prompt templates, narrative structures, privacy settings — in a single selection.
Documentary
Emotional depth. Authentic moments. Cultural sensitivity.
Selects for
Emotional weight, thematic resonance, pauses that carry meaning
Narrative
Emergent — arc follows the material, not a template
Privacy
Local only. Cultural content flagging available.
Brand
Message clarity. Stakeholder alignment. Brief discipline.
Selects for
Clarity, confidence, conciseness, alignment with creative brief
Narrative
Brief-driven — structure follows brand pillars and key messages
Privacy
Cloud LLMs enabled for faster, higher-quality analysis
Commercial Doc
Real stories. Real people. Real marketing framework.
Selects for
Authenticity balanced with message discipline and tone alignment
Narrative
Hybrid — emergent themes shaped by messaging requirements
Privacy
Project-dependent. You choose per engagement.
Under the Hood
Built for Apple Silicon
Transcription
faster-whisper
Native M-series Neural Engine. 60-min interview in under 4 minutes.
Audio Analysis
librosa + numpy
RMS energy, pitch contour, speech rate, spectral features.
LLM Layer
Ollama / Claude
Local 70B models or cloud API. litellm unified interface.
Timeline Export
EDL / FCPXML
Frame-accurate. DaVinci Resolve Studio native import.
Workflow
Command line at the core
Every pipeline stage is one command. Run them individually, or let plotline run handle the full sequence. HTML reports open in your browser for review.
$ plotline init elder-series --profile documentary
✓ Project created with documentary profile
 
$ plotline add interviews/*.mov
✓ 6 interviews added (4h 12m total)
 
$ plotline run
Transcribing .......... done (18m 34s)
Analyzing audio ....... done (6m 12s)
Extracting themes ..... done (8m 47s)
Cross-referencing ..... done (3m 22s)
Building arc .......... done (5m 11s)
 
→ 47 segments selected across 4 themes
→ Review report: open in browser? (Y/n)
"
The gap between watching ten hours of interviews and building a ten-minute story is where documentary lives. Plotline doesn't close that gap — it illuminates it.
Michael Warf — Director, Coalbanks Creative Inc.
Stop scrubbing.
Start cutting.

Plotline is free and open source. Clone the repo and start transcribing your interviews in minutes.

View on GitHub → Read the Docs

Free and open source. Python. Local-first. Your footage stays on your machine.