The Voice Notebook

“Whose speaking style is magnetic—or infuriating?”

This notebook captures vocabulary, rhythm, and audience vibe from online personalities, influencers, podcasters, or public figures. It’s raw material for dialogue, narrative voice, and character authenticity.

Notebook Name: Raw – Voice – [Handle/Person Name]

Ready to extract? After collecting your sources, use the Deepening Prompts to create structured ingredients for Writers Factory.


Source Checklist

Goal: Capture prose rhythm, diction, and narrative stance for VOICE and CHARACTER work.

Include (3–5 items):

  • 2–3 short textual excerpts (1–3 pages each) Passages where the voice is pure and unmistakable.
  • 1 audio or transcript An interview, talk, or podcast where you hear the author or narrator voice.
  • 1 “Reading Trace / Voice Notes” document Where you mark sentence length patterns, recurring words, typical metaphors, dialogue quirks.

Avoid:

  • Full collected works or whole novels in one notebook.
  • Mixing many different authors’ voices in the same notebook.

Why This Notebook Matters

Social media and podcasts have created a new landscape of public speech. The people who succeed online have developed distinctive voices—patterns of word choice, rhythm, and emotional register that cut through noise.

Whether you find them compelling or repellent, studying these voices teaches you:

  • How real people actually talk (vs. literary dialogue)
  • Generational/cultural vocabulary (slang, references, shibboleths)
  • Audience psychology (what makes people engage)

You don’t have to like them. You have to understand why they work.


What to Collect

1. Posts & Threads (Sparks)

Quantity: 10–30 posts, grouped by theme

Source Type Where to Find It File Format
Twitter/X threads Copy into document, or screenshot TXT, PNG
Substack posts Export or PDF save PDF, TXT
YouTube video descriptions Copy text TXT
LinkedIn posts Screenshot or copy PNG, TXT
Instagram captions Screenshot PNG

How to Capture Threads:

  • Use Thread Reader to unroll Twitter threads to PDF
  • Screenshot carousel posts with all slides
  • Copy full post text including line breaks (formatting matters)

2. Long-Form Content (Depth)

Quantity: 1–2 pieces

Source Type Where to Find It File Format
Podcast episodes Download from app or RSS MP3
YouTube videos Rip audio or use auto-transcript MP3, TXT
Newsletter issues Export from email or Substack PDF
Interviews Search “[Name] interview podcast” MP3

Why Audio Matters: Hearing someone speak reveals cadence, pauses, verbal tics—the rhythm beneath the words.


3. The Vocabulary List (Rhythm)

Quantity: 1 document, 10–20 terms

Create a document listing words, phrases, and speech patterns unique to this voice:

VOCABULARY LIST: [Person/Handle]

SIGNATURE PHRASES:
- [Phrase they use constantly]
- [Catchphrase or tagline]
- [Unique formulations]

SLANG/JARGON:
- [Terms their audience uses]
- [In-group vocabulary]
- [Abbreviations, acronyms]

VERBAL TICS:
- [Filler words: "like," "right?", "look—"]
- [Sentence starters: "Here's the thing..."]
- [Rhetorical patterns: frequent questions? Lists?]

FORBIDDEN WORDS:
- [What they NEVER say—revealing omissions]

4. Your Analysis (Why It Works)

Quantity: 1 document

Don’t just collect—analyze. What makes this voice effective (or annoying)?

VOICE ANALYSIS: [Person/Handle]

THE HOOK: What grabs attention in the first 3 seconds?

THE RHYTHM: Short sentences? Long sprawling thoughts? Mix?

THE STANCE: Are they teaching? Ranting? Confiding? Performing?

THE AUDIENCE: Who are they talking TO? (Not everyone—specifically who?)

WHAT MAKES IT WORK:
[Why do people engage? What need does this voice fill?]

WHAT'S ANNOYING ABOUT IT:
[Be honest. What grates? This is equally useful data.]

CHARACTERS WHO COULD SPEAK LIKE THIS:
[Who in your story (or future story) might use this voice?]

Source-Hunting Strategies

For Twitter/X Voices

  1. Search for viral threads by handle
  2. Use Thread Reader for easy export
  3. Look at “Top” posts filtered by timeframe
  4. Check their “Likes” for what resonates with them

For YouTube/Podcast Voices

  1. Start with their most-viewed video (highest signal)
  2. Find interview appearances where they’re less polished
  3. Download auto-generated transcripts (Settings > Transcripts)
  4. Descript can transcribe audio files for free

For Newsletter/Substack Voices

  1. Read their “Best of” or pinned posts first
  2. Archive pages often have full backlist
  3. Forward to yourself, save as PDF
  4. Note their subject line patterns (that’s voice too)

For Public Figures (Politicians, CEOs, etc.)

  1. Speeches: YouTube has most major speeches
  2. Transcripts: News sites often publish full text
  3. Press conferences: Less scripted, more revealing
  4. Memoirs/Ghostwritten books: Voice coaching made explicit

Organizing Your Sources (Optional)

You can use prefixes to organize your raw sources during collection:

Prefix Type Example
[THREAD] Twitter/social thread [THREAD] - @pmarca on tech optimism
[POST] Single post or article [POST] - Naval Ravikant - How to Get Rich
[EPISODE] Podcast/video [EPISODE] - Lex Fridman #287

Note: These organize your raw sources. The Deepening Prompts handle extracting structured ingredients—no special naming required for those outputs.


Create a Saved Note titled: WHY I CHOSE THIS VOICE

[!IMPORTANT] Don’t Forget: After writing this note, select it and click “Convert to Source”. The AI cannot read your Saved Notes unless they are converted into Sources!

Template:

PERSON/HANDLE: [Name]

WHAT MAKES THIS VOICE MAGNETIC (OR ANNOYING):
[2–3 sentences on why you're studying them]

WHAT I'M LOOKING FOR:
- [ ] Vocabulary/slang (specific word choices)
- [ ] Rhythm/cadence (sentence patterns)
- [ ] Rhetorical moves (how they persuade)
- [ ] Audience relationship (how they create intimacy/authority)
- [ ] Anti-patterns (what to AVOID)

CHARACTER APPLICATION:
[Which character type in fiction could speak like this?]
[Or: What cultural moment does this voice represent?]

Example:

PERSON/HANDLE: @dril (Twitter absurdist)

WHAT MAKES THIS VOICE MAGNETIC:
Surrealist comedy delivered with complete deadpan commitment.
The humor comes from the gap between mundane format (tweet)
and unhinged content. It's a masterclass in voice-as-character.

WHAT I'M LOOKING FOR:
- [x] Vocabulary/slang (weirdly formal phrases for absurd claims)
- [x] Rhythm/cadence (short, declarative, no hedging)
- [ ] Audience relationship
- [x] Anti-patterns (what makes parodies fail?)

CHARACTER APPLICATION:
A character who speaks with total confidence about things that make
no sense. The unhinged coworker. The cult leader. The unreliable
narrator who doesn't know they're unreliable.

Common Mistakes

Mistake Problem Fix
Only collecting people you agree with Misses range Include voices that annoy you—they’re equally useful
Ignoring the audience Voice is a relationship Ask: who are they talking TO?
Surface-level collection No analytical value Write the vocabulary list and voice analysis
Too many voices Muddy signal 1 voice per notebook, go deep
Forgetting audio Missing rhythm Always include at least 1 audio source

Example Notebook Structure

Raw – Voice – Paul Graham/
├── [THREAD] - How to Do Great Work (thread copy).txt
├── [POST] - The Top Idea in Your Mind.pdf
├── [POST] - Maker's Schedule, Manager's Schedule.pdf
├── [EPISODE] - Lex Fridman Interview.mp3
├── [VOCAB] - PG Vocabulary Analysis.txt
├── Voice Analysis - Paul Graham.txt
└── WHY I CHOSE THIS VOICE (Saved Note)

What This Feeds Into

When you run the Deepening Prompts, this notebook produces:

Table What You’ll Get
voice_analysis POV type, sentence rhythm, anti-patterns, sample techniques
character_archetypes Dialogue patterns, psychological reveals

← Visual Language Next: The Arena →