For Developers

Local dictation for
agentic coding workflows

Press a key, speak naturally, clean formatted text appears in Claude Code, Cursor, VS Code, or any terminal — transcribed and edited on your device, never the cloud.

Download for macOS

Free trial — 3,000 words or 7 days, no account required.

Drive Claude Code, Cursor, and terminals by voice

Prompt your AI coding assistants, run shell commands, write commit messages, and draft code comments — all without touching the keyboard. evoglyph delivers clean, formatted text directly into the focused field using the right injection strategy for each app.

Native Cocoa editors

Xcode, TextEdit, Notes, and similar native macOS apps use AXUIElement read-splice-write. Text lands at the caret with no clipboard side effects and full undo history.

See Text injection for the full strategy table.

VS Code, Cursor & Electron apps

Electron-based editors and web-rendered surfaces receive text via Pasteboard + Cmd+V — evoglyph detects the app type automatically and routes accordingly. Cursor and VS Code autocomplete popups can be dismissed with Escape before dictating.

Per-app notes at Target apps.

Terminals (iTerm2 & others)

Terminal-class apps receive text via Pasteboard + Cmd+V. Note: iTerm2's Secure Keyboard Entry setting blocks all injection — disable it in iTerm2 › Secure Keyboard Entry while using evoglyph.

Details at Target apps.

The injection strategy is automatic — there's nothing to configure per app. For the full routing table and per-app notes, see Text injection and Target apps.

Developer vocabulary — gRPC, OAuth, Postgres

The base model is general-purpose. Add your tech stack's jargon once and evoglyph biases the decode toward those terms across every dictation.

  • CTC keyword boosting increases the probability of your terms winning the decode — no model retraining, no per-user fine-tuning
  • Handles acronyms, camelCase, and compound names: WebSocket, TypeScript, SwiftUI, Kubernetes
  • Changes take effect on the next dictation — no restart, no apply step
  • The vocabulary list and all scoring runs on-device — nothing leaves your Mac

Custom vocabulary docs →

Example: tech stack terms

  • gRPC “grumpy C” → gRPC
  • Postgres “prosegrass” → Postgres
  • OAuth “oh auth” → OAuth
  • WebSocket ambiguous → WebSocket

Examples from docs/vocabulary.html — the vocabulary docs page.

Per-app setup

evoglyph routes text injection automatically based on app type. The Target apps page covers every supported app with the exact injection path, known gotchas, and per-app notes for Claude desktop, iTerm2, VS Code, Cursor, Slack, and more.

View the target-apps guide →

Fast enough to keep pace with thought

evoglyph's speech-to-text engine-time (the time the transcription model takes from end-of-speech to decoded text) is measured at 85 ms at the p50 on Apple Silicon. That's the engine step alone; total time from hotkey release to text appearing includes cleanup latency on top.

Figures are engine-time measurements from internal benchmarks; methodology is available on request. For the full numbers, methodology notes, and accuracy results (2.41% WER on LibriSpeech test-clean), see /benchmark.

85 ms

p50 engine-time

124 ms

p95 engine-time

2.41%

WER (LibriSpeech test-clean, our harness)

Engine-time: time from end-of-speech to decoded text (transcription model only). Figures from internal eval results, 2026-05-11; methodology available on request via [email protected].

Your code never leaves the machine

Every transcription and AI cleanup step runs on your Mac. Audio is captured, processed, and discarded locally — no audio, no text, and no prompts are ever transmitted to any server.

  • Speech is transcribed on-device by Parakeet TDT v2 running on the Apple Neural Engine
  • AI cleanup runs on-device — the cleanup model is downloaded once at first launch and stays local
  • No analytics, no telemetry by default — crash reports are opt-in only
  • Dictation history stays in a local SQLite database on your disk, never synced to a server

Full network event table →

The four network calls evoglyph makes

  • 1. Model download HuggingFace, first launch only. Never again after that.
  • 2. License validation LemonSqueezy, periodic re-check. No audio involved.
  • 3. Update check Sparkle against evoglyph.com. Version numbers only.
  • 4. Crash reports Sentry, opt-in only — off by default in Settings › Diagnostics.

Audio never leaves your device. See where-your-audio-goes.html for the full breakdown.

Start dictating into your dev tools

Requires Apple Silicon and macOS 14 (Sonoma) or later. No account needed to start.

Download for macOS

Free trial — 3,000 words or 7 days, no account required.
macOS 14+ • Apple Silicon