Obsidian Meeting Notes Complete Guide

Introduction: The Meeting Problem in Personal Knowledge Management

Obsidian is renowned for turning scattered thoughts into resilient personal knowledge graphs. By storing notes as durable Markdown files on your local drive, it frees users from proprietary database lock-in. Yet, for knowledge workers, product managers, and engineering leaders, an enormous fraction of intellectual work does not originate from quiet reading—it happens in real-time conversations across video meetings.

Unfortunately, integrating verbal meetings into an Obsidian vault has historically been fraught with friction. Users either scramble to type live notes during fast-paced calls, record voice memos on mobile phones that must be manually transcribed later, or stitch together fragile community plugins that require cloud API keys and monthly subscriptions.

In this comprehensive guide, we unpack the optimal architecture for managing meeting notes in an Obsidian vault. We cover folder taxonomies, YAML frontmatter schemas, automated botless audio capture on macOS, on-device Whisper transcription, and knowledge graph backlink integration.

---

1. Vault Taxonomy: Structuring Meeting Notes for Decadal Recall

A meeting note should never exist as an isolated, orphan document. To ensure that discussions connect to ongoing projects, customer relationships, and strategic roadmaps, your vault needs a structured taxonomic foundation.

1.1 The Directory Layout

We recommend maintaining a dedicated top-level directory for all conversational records:

MyVault/
├── 00-Inbox/
├── 10-Projects/
│   └── Phoenix/
├── 20-Areas/
├── 30-People/
│   ├── Marcus Vance.md
│   └── Helena Vance.md
├── 40-Meetings/
│   ├── 2026/
│   │   ├── 10-October/
│   │   │   ├── 2026-10-09-phoenix-architecture-review.md
│   │   │   └── 2026-10-09-marcus-weekly-one-on-one.md
└── 90-Templates/
    └── meeting-template.md

By enforcing the ISO 8601 naming format (YYYY-MM-DD-topic-slug.md), notes sort chronologically in standard operating system file browsers, preserving timeline context even outside Obsidian.

1.2 Essential YAML Frontmatter Schema

Every meeting note should start with standardized YAML metadata. This allows Obsidian Dataview queries to aggregate action items across months of discussions:

---
title: "Phoenix Architecture & Storage Review"
date: 2026-10-09
meeting_type: "Architecture Review"
attendees: ["[[Marcus Vance]]", "[[Helena Vance]]", "[[David Chen]]"]
project: "[[Project Phoenix]]"
status: "Completed"
audio_file: "attachments/audio-2026-10-09-phoenix.m4a"
tags: [meeting, architecture, storage, phoenix]
---

Using wikilinks inside frontmatter lists ("[[Marcus Vance]]") ensures that Obsidian’s metadata cache indexes participant connections immediately.

---

2. Capturing Meeting Audio: The Botless macOS Pipeline

The most common failure mode in meeting documentation is typing fatigue. When you focus on typing verbatim quotes, your cognitive presence in the strategic conversation diminishes.

2.1 The Limitations of Cloud Recording Bots

Traditional meeting tools deploy virtual bots (OtterPilot, Fred, etc.) that join your call as guest accounts. In professional environments, bots introduce multiple drawbacks:

• Client Discomfort: External clients and legal partners often object to unvetted cloud bots recording confidential deliberations.
• Join Delays: Bots frequently get trapped in meeting waiting rooms, missing the first few minutes of critical context.
• Data Exposure: Audio is streamed to third-party cloud data centers, creating regulatory compliance concerns.

2.2 Native Loopback Audio via ScreenCaptureKit

A modern local-first workflow bypasses cloud bots entirely by leveraging Apple’s ScreenCaptureKit framework on macOS 14.2+.

Local Botless Capture Pipeline:
[macOS System Audio (Zoom/Meet)] ──┐
                                   ├──> [ScreenCaptureKit Engine] ──> [Local Ring Buffer]
[Hardware Microphone Input]      ──┘                                        │
                                                                       (Local Core ML)
                                                                            ▼
                                                                [Whisper Speech-to-Text]
                                                                            ▼
                                                            [Obsidian Vault: meeting.md]

Using Oyma, meeting audio from both your microphone and your call attendees is captured quietly at the operating system level, transcribed locally on Apple Silicon, and saved directly as Markdown inside your target Obsidian folder.

---

3. On-Device Speech Recognition with Local Whisper

Once audio is captured, speech recognition transforms raw waveforms into timestamped text. Rather than transmitting client audio to cloud APIs, modern Apple Silicon hardware allows you to run OpenAI Whisper locally.

3.1 Model Selection for Meeting Notes

Depending on your Mac’s hardware specifications, select the model that balances speed and precision:

• Whisper Base (148 MB): Near-instantaneous inference on M1 Macs; suitable for clear single-speaker audio.
• Whisper Small (488 MB): Excellent balance of speed and multi-speaker conversational precision.
• Whisper Large v3 Turbo (1.6 GB): State-of-the-art accuracy across technical jargon, foreign accents, and noisy environments. Runs at up to 12.5x real-time speed on an M3 Max.

3.2 Clickable Timestamp Anchors

A high-fidelity meeting note should link text directly back to the spoken word. In Oyma, meeting transcripts generate clickable Markdown timestamp anchors:

- `[04:12]` Marcus Vance outlined the replication lag benchmarks for SQLite.
- `[08:45]` Helena Vance confirmed that local FileVault encryption satisfies client compliance.

Clicking [04:12] seeks directly to the exact second in the local audio recording, allowing instant verification without scrubbing through an hour of tape.

---

4. Local AI Summarization: Formatting with Ollama

A raw 60-minute meeting transcript can easily span 9,000 words. For knowledge recall, a structured executive summary is essential.

4.1 Running Local LLMs Offline

By pairing Oyma with Ollama, meeting transcripts can be synthesized on your Mac using open-weights models like Llama 3.1 8B or Mistral Nemo:

• Zero Cloud Costs: No recurring API token fees.
• Zero Data Egress: Summarization executes entirely in unified memory.
• Consistent Markdown Output: Structured templates enforce bulleted decisions and actionable tasks.

4.2 Standard Meeting Note Scaffold

The generated meeting note lands directly in your Obsidian vault with the following clean anatomy:

# Phoenix Architecture & Storage Review

## 1. Executive Summary
Marcus and Helena reviewed storage engine options for the local-first client. The team approved adopting plain CommonMark files as the primary source of truth, utilizing SQLite FTS5 exclusively as an ephemeral search cache.

## 2. Strategic Decisions
- **Decision 1:** Reject proprietary binary database storage to ensure 50-year document durability.
- **Decision 2:** Implement an atomic file-writing pipeline with pre-write backup scratchpads to prevent corruption during unexpected shutdowns.

## 3. Action Items & Owners
- [ ] [[Marcus Vance]] Benchmark read replica replication lag under synthetic load by Wednesday.
- [ ] [[David Chen]] Finalize backward-compatible database adapter layer for legacy services.
- [ ] [[Helena Vance]] Review client audit checklist ahead of Friday steering committee.

## 4. Raw Transcript Highlights
- `[02:15]` Team aligned on quarterly performance metrics.
- `[15:40]` Debate regarding memory allocations on 8GB unified memory Mac models.

---

5. Integrating Notes with the Obsidian Knowledge Graph

The true power of capturing meetings in Obsidian emerges after the call ends, as new notes weave into your broader personal knowledge graph.

5.1 Person MOCs (Maps of Content)

When attendees are wikilinked ([[Marcus Vance]]), navigating to Marcus’s note instantly reveals every meeting he attended via Obsidian’s Backlinks panel. You can quickly review past commitments, shared decisions, and discussion histories before entering your next 1-on-1.

5.2 Dataview Query for Global Action Items

By embedding task checkboxes (- [ ]), you can create a central “Action Items Dashboard” in your daily note or project dashboard using the Dataview plugin:

TASK
FROM "40-Meetings"
WHERE !completed
SORT file.name DESC

This query dynamically pulls every open commitment across all meeting notes into a single unified task list, ensuring no client deliverable slips through the cracks.

---

6. Practical Setup: Connecting Oyma to Obsidian

Connecting Oyma to your existing Obsidian vault takes less than two minutes:

1. Open Oyma Settings (Cmd + ,).
2. Select Storage & Vaults.
3. Under Vault Destination, click Choose Directory and select your Obsidian vault folder (e.g., ~/Documents/ObsidianVault/Meetings/).
4. Select your preferred default meeting template from our Markdown templates catalog.
5. All future meeting recordings will automatically save structured, wikilinked .md files directly into your vault folder with zero cloud sync required.

---

7. Conclusion & Next Steps

Meeting documentation no longer requires choosing between stressful live typing and risky cloud recording bots. By marrying Obsidian’s durable local-first file architecture with on-device speech intelligence, knowledge workers can capture every critical nuance of their conversations with complete data sovereignty.

To learn more, explore our vault health inspector tool, read our analysis of the real cost of AI in Obsidian, or download our free meeting note templates.

---

8. Visualizing Meeting Networks with Obsidian Canvas

Beyond linear notes and Dataview tables, complex product initiatives and architectural design reviews benefit enormously from visual synthesis using Obsidian Canvas.

Obsidian Canvas Sprint Topology:
┌─────────────────────────┐          ┌─────────────────────────┐
│ [[2026-10-09 Sprint]]   │─────────►│ [[ADR-042 RocksDB]]     │
│ Note Card (Summary)     │          │ Architecture Decision   │
└───────────┬─────────────┘          └─────────────────────────┘
            │
            ▼
┌─────────────────────────┐          ┌─────────────────────────┐
│ [[Julian Thorne]]       │─────────►│ Action Items Checklist  │
│ Participant Card        │          │ Embedded Live Tasks     │
└─────────────────────────┘          └─────────────────────────┘

8.1 Constructing Multi-Dimensional Sprint Reviews

1. Drop Meeting Notes Directly: Drag the generated meeting note from your file explorer into an active .canvas board.
2. Embed Action Item Callouts: Use transclusion (![[2026-10-09-meeting#Action Items]]) to embed live checklists directly within visual project roadmaps.
3. Connect Decision Nodes: Draw directional arrows connecting participant contributions to architectural decision records (ADRs), establishing transparent accountability for technical choices.

---

8. Visualizing Meeting Knowledge with Obsidian Canvas

In addition to linear Markdown files and Dataview tables, complex multi-stakeholder architectural syncs benefit immensely from visual knowledge mapping using Obsidian Canvas.

Obsidian Canvas Meeting Topology:
┌───────────────────────────┐           ┌───────────────────────────┐
│ [[2026-10-09 Sprint Mtg]] │──────────►│ [[ADR-042 RocksDB FTS5]]  │
│ Primary Meeting Card      │           │ Architectural Decision    │
└─────────────┬─────────────┘           └───────────────────────────┘
              │
              ▼
┌───────────────────────────┐           ┌───────────────────────────┐
│ [[Julian Thorne]]         │──────────►│ Open Deliverables List    │
│ Participant Profile       │           │ Live Embedded Checklists  │
└───────────────────────────┘           └───────────────────────────┘

8.1 Constructing Visual Architecture Review Boards

1. Drag and Drop Meeting Notes: Drag newly generated meeting files directly onto an open .canvas board.
2. Transclude Live Task Blocks: Use Markdown transclusion (![[2026-10-09-sync#Action Items]]) to render live, interactive checkboxes inside the visual topology.
3. Connect Decision Nodes: Draw directional connection lines from participant cards to Architectural Decision Records (ADRs), establishing transparent historical lineage for architectural choices.

---

9. Automated Sprint Rollups with Embedded DataviewJS

Beyond manual canvas visualization, engineering leads can aggregate action items across dozens of distributed standup notes using embedded JavaScript queries:

// Render an interactive table of uncompleted engineering tasks across all recent meetings
const syncNotes = dv.pages('"40-Meetings"')
  .where(p => p.type === "meeting-note" && p.tags?.includes("sprint"))
  .sort(p => p.date, "desc");

const pendingTasks = [];
for (const note of syncNotes) {
  const file = app.vault.getAbstractFileByPath(note.file.path);
  const content = await app.vault.read(file);
  const taskRegex = /^- \[ \] (.*?)(\[\[.*?\]\])?$/gm;
  let match;
  while ((match = taskRegex.exec(content)) !== null) {
    pendingTasks.push([note.date, note.file.link, match[1].trim()]);
  }
}

dv.table(["Meeting Date", "Note Link", "Unfinished Deliverable"], pendingTasks);

By embedding this script into your team’s weekly sprint dashboard note, every newly recorded action item surfaces automatically without manual status copying.