Technical Glossary · Computing & Privacy
The local-first computing & meeting notes glossary.
Clear, rigorous definitions for modern personal knowledge management, on-device Whisper transcription, Markdown specifications, and data sovereignty.
Every definition includes technical specifications, architectural trade-offs, standards citations, and native implementation details in Oyma.
Understanding the Oyma architectural vocabulary
Modern desktop software is undergoing an architectural renaissance. Users and security-sensitive organizations are shifting away from centralized cloud databases toward local-first architectures, private on-device intelligence, and durable plain-text files. This glossary provides exact technical specifications for the concepts, standards, and algorithms underpinning Oyma on macOS.
Architecture & Storage
15 termsLocal-First Software
A software architecture paradigm where applications prioritize reading, writing, and persisting data directly to the user's local hardware before any network operations.
Read technical guide →
Data Sovereignty
The legal, philosophical, and architectural principle that an individual or organization maintains exclusive ownership, control, and jurisdictional governance over their digital information.
Read technical guide →
Plain Text Computing
A computing discipline where unstructured or lightly structured human-readable text encoded in UTF-8 serves as the permanent format for notes, documentation, and data.
Read technical guide →
File Over App Architecture
An architectural philosophy asserting that individual files and folders on the filesystem should be the primary interface and permanent asset, with software apps serving merely as interchangeable tools.
Read technical guide →
CommonMark Specification
A formal, unambiguous, and mathematically rigorous specification and test suite for the Markdown markup language, maintained by the CommonMark working group.
Read technical guide →
GitHub Flavored Markdown (GFM)
A formal extension of the CommonMark specification created by GitHub, standardizing tables, task checkboxes, strikethrough, and autolinks.
Read technical guide →
Obsidian Flavored Markdown
An extended Markdown dialect popularized by Obsidian that adds bidirectional wikilinks, callout admonitions, math expressions, and frontmatter properties.
Read technical guide →
Bidirectional WikiLinks
A compact double-bracket syntax ([[Target Note]]) that creates an associative hyperlink between two documents without requiring explicit file paths or extensions.
Read technical guide →
Backlinks in Note Taking
An incoming link pointing to a document from another file, revealing where, why, and how a concept has been referenced across an entire knowledge repository.
Read technical guide →
YAML Frontmatter
A metadata block delimited by triple hyphens (---) at the very top of a Markdown document that defines structured key-value attributes for parsers and databases.
Read technical guide →
SQLite FTS5 Full-Text Search
An embedded virtual table module for the SQLite database engine that compiles documents into an inverted index for lightning-fast keyword and phrase retrieval.
Read technical guide →
BM25 Ranking Algorithm
A state-of-the-art probabilistic information retrieval function that ranks matching documents based on term frequency, inverse document frequency, and document length normalization.
Read technical guide →
Zero Vendor Lock-In
A software design philosophy guaranteeing that users can export, inspect, migrate, or utilize their data with competing tools at any time with zero technical friction or formatting loss.
Read technical guide →
macOS Application Sandbox
An Apple operating system security technology that isolates applications within tightly restricted containers, restricting access to system hardware, network sockets, and user files.
Read technical guide →
Atomic Notes Principle
A note-taking principle stipulating that each document should encapsulate exactly one discrete idea, concept, or insight, making it modular and linkable.
Read technical guide →
Audio & Machine Learning
15 termsAutomatic Speech Recognition (ASR)
A computer science and machine learning discipline that converts human acoustic speech waveforms into machine-readable digital text tokens in real time.
Read technical guide →
OpenAI Whisper Model
An open-weights, end-to-end encoder-decoder transformer neural network architecture trained on 680,000 hours of weakly supervised audio for multilingual speech recognition.
Read technical guide →
Apple CoreAudio Framework
Apple's native low-level digital audio subsystem and API framework for macOS and iOS, designed for high-throughput, sub-millisecond audio streaming and processing.
Read technical guide →
ScreenCaptureKit Audio Pipeline
An Apple framework introduced in macOS 12.3 that provides hardware-accelerated, low-overhead capture of display streams and application system audio with explicit user permissions.
Read technical guide →
Loopback Audio Routing
A software or hardware routing mechanism that redirects an operating system's speaker output stream directly back into an audio input stream for recording or analysis.
Read technical guide →
Audio Buffering in Speech Processing
A dedicated region of physical memory (RAM) used to temporarily hold incoming digitized audio samples before they are processed by an ASR neural network.
Read technical guide →
Neural Network Quantization
A model optimization technique that compresses deep learning weights from high-precision floating-point values into lower-bit integers to reduce memory and accelerate execution.
Read technical guide →
Apple Neural Engine (ANE)
A specialized neural processing unit (NPU) integrated into Apple Silicon SoCs (M-series and A-series chips) dedicated to energy-efficient tensor operations.
Read technical guide →
Metal Performance Shaders (MPS)
A collection of highly tuned GPU compute and graphics shader primitives provided by Apple to accelerate linear algebra, neural networks, and image processing on macOS.
Read technical guide →
Large Language Model Hallucination
A phenomenon where an artificial intelligence language model generates syntactically fluent and confident statements that are factually false, ungrounded, or unsupported by source data.
Read technical guide →
Context Window in Language Models
The maximum quantity of tokenized information (prompt plus response) that a large language model can hold in active memory and process during a single inference cycle.
Read technical guide →
Prompt Engineering for Summaries
The disciplined engineering practice of structuring, constraining, and refining natural language inputs to guide large language models toward deterministic, reliable outputs.
Read technical guide →
On-Device AI Inference
The execution of trained artificial intelligence and machine learning models directly on the client user's physical hardware without transmitting data to remote cloud servers.
Read technical guide →
Word Error Rate (WER)
The universal benchmark metric used to evaluate automatic speech recognition accuracy, calculated by summing word substitutions, deletions, and insertions divided by reference word count.
Read technical guide →
Voice Activity Detection (VAD)
An audio signal processing technique that detects the presence or absence of human speech in an audio stream, filtering out background silence and noise.
Read technical guide →
Privacy & Security
15 termsZero Telemetry Architecture
A privacy engineering standard guaranteeing that an application transmits zero analytical, diagnostic, behavioral, or heartbeat payloads to remote vendor servers.
Read technical guide →
Outbound Egress Monitoring
The practice of tracking, logging, and inspecting all network traffic initiated by local applications attempting to communicate with external internet servers.
Read technical guide →
Botless Meeting Recording
A meeting audio capture architecture that records conference sound directly from the local computer's audio output without injecting a bot participant into the call.
Read technical guide →
Air-Gapped Computing
A computer security posture where a device or application operates in complete physical and electronic isolation from external computer networks and the public internet.
Read technical guide →
Client-Side Encryption
A cryptographic security model where data is encrypted directly on the client user's device using private keys before it is stored or transmitted, ensuring third parties cannot read it.
Read technical guide →
SOC 2 Type II Security Standard
An independent auditing framework developed by the AICPA that evaluates an organization's controls governing security, availability, processing integrity, confidentiality, and privacy.
Read technical guide →
HIPAA Compliance in Digital Health
A United States federal statutory framework that mandates strict data privacy and security provisions for safeguarding Protected Health Information (PHI).
Read technical guide →
GDPR Article 17 Right to Erasure
A fundamental European Union privacy regulation granting individuals the legal right to request the complete deletion of their personal data without undue delay.
Read technical guide →
Data Residency Regulations
The legal and regulatory requirement that digital information generated within a specific jurisdiction must remain physically stored and processed within those borders.
Read technical guide →
Two-Party Consent Call Recording
A legal requirement in certain jurisdictions mandating that all participants in a communication must be notified and give explicit consent before a conversation can be recorded.
Read technical guide →
Differential Privacy Mathematical Model
A formal mathematical framework that quantifies and bounds the individual privacy risk incurred when sharing statistical analyses or aggregate datasets.
Read technical guide →
Plain Text Storage at Rest
An architectural choice where files are stored as standard, unencrypted UTF-8 text on local disk, delegating cryptographic security to whole-volume disk encryption.
Read technical guide →
Apple FileVault Encryption
Apple's native full-disk volume encryption system for macOS, securing all user data on storage media using hardware-accelerated XTS-AES-128 encryption.
Read technical guide →
Bring Your Own Key (BYOK)
A software architecture model where users supply their own private API credentials (e.g., OpenAI, Anthropic) directly to an application, eliminating middleman proxies and markups.
Read technical guide →
Confidential Hardware Computing
A security paradigm that protects sensitive data during active processing by isolating computation inside a hardware-attested Trusted Execution Environment (TEE).
Read technical guide →
Knowledge & Productivity
15 termsZettelkasten Note-Taking Method
A systematic personal knowledge management methodology based on interconnected, atomic index cards organized through alphanumeric identifiers and cross-references.
Read technical guide →
Building a Second Brain Methodology
A digital personal knowledge management methodology designed to systematically capture, organize, distill, and express intellectual assets using digital tools.
Read technical guide →
The P.A.R.A. Organization Method
A universal digital organization framework developed by Tiago Forte that categorizes all notes and files into four actionable folders: Projects, Areas, Resources, and Archives.
Read technical guide →
Bidirectional Linking Architecture
A hyperlinking model where creating a forward link from document A to document B automatically generates a reciprocal, queryable backlink from B to A.
Read technical guide →
Knowledge Graph Visualization
A visual network model and interactive map that represents individual notes as nodes and their hyperlinks as connecting edges, revealing intellectual clusters and relationships.
Read technical guide →
Daily Notes Workflows
A time-anchored note-taking practice where a dedicated document is automatically generated for each calendar day, serving as a friction-free inbox and journal.
Read technical guide →
Action Items Extraction
Discrete, actionable tasks assigned to specific individuals with unambiguous deliverables and deadlines, extracted from meeting discussions.
Read technical guide →
Formal Meeting Minutes Protocol
An official, structured written record documenting attendees, proceedings, motions, votes, and resolutions of a formal corporate, board, or legislative meeting.
Read technical guide →
Blameless Post-Mortem Architecture
An engineering incident investigation practice focused on uncovering systemic and organizational root causes without assigning personal fault or retribution.
Read technical guide →
Request for Comments (RFC) Process
A formal engineering proposal document that outlines a technical architecture, protocol, or system change to solicit feedback and achieve consensus before implementation.
Read technical guide →
Product Requirements Document (PRD)
A foundational product management document that articulates the purpose, scope, user personas, functional specifications, and success criteria for a software release.
Read technical guide →
Live-Preview Markdown Editing
An editor interface architecture that renders Markdown syntax as rich typography in real time while revealing raw plain-text delimiters when the cursor enters an active line.
Read technical guide →
Outliner Note-Taking Hierarchy
A note-taking interface where all text is structured as a hierarchical tree of collapsible, indented bullet points with individual block identity.
Read technical guide →
Markdown Callout Admonitions
A visual admonition container styled with accent borders and icons, authored in Markdown using blockquote syntax paired with an archetype identifier.
Read technical guide →
Document Transclusion Embeds
The dynamic inclusion of content from one source document directly inside another file by reference, displaying live updates without duplicating text.
Read technical guide →
Why terminology precision matters in personal computing
Marketing buzzwords frequently obscure critical software trade-offs. Terms like "cloud sync", "AI assistant", and "automatic transcription" often conceal hidden security compromises, such as unencrypted audio storage on third-party servers, training machine learning models on private meeting transcripts, and proprietary database schemas that prevent data portability.
At Oyma, we prioritize engineering clarity over ambiguity. We distinguish between true local-first software and cloud-dependent desktop shells; between transparent plain UTF-8 text and proprietary document bundles; and between zero-bot system audio capture via ScreenCaptureKit and cloud bots that intrude on executive conversations.
Explore related technical documentation: read about our zero-telemetry privacy architecture, browse our Markdown syntax reference, inspect note structures with the Vault Health Inspector, or examine local AI with Ollama.
Run private Markdown notes on your Mac.
Local-first storage, on-device Whisper transcription, and zero cloud telemetry.