How does dIKta.me work offline?

dIKta.me runs Whisper V3 Turbo and local LLMs (Gemma 3, Llama 3) directly on your GPU. No audio or text data ever leaves your machine. It is 100% air-gapped by default.

What operating systems does dIKta.me support?

dIKta.me is available for Windows 10+ (x64). macOS and Linux support are on the roadmap.

How much does dIKta.me cost?

Free trial with cloud credits included. Full Version: $20 one-time purchase for unlimited local dictation, all voice modes, and lifetime updates. No subscription required.

What languages does dIKta.me support for speech recognition?

Whisper V3 Turbo supports 90+ languages with automatic language detection. Bidirectional English-Spanish translation is built-in.

Do I need an NVIDIA GPU to use dIKta.me?

An NVIDIA GPU is recommended for the fastest local STT and LLM processing. However, dIKta.me also works on CPU (slower) and offers a cloud mode with wallet credits for users without a powerful GPU.

LOCAL. PRIVATE. NO SUBSCRIPTIONS.

STOP TYPING.
START
TALKING.
THINKING.
WORKING.
WINNING.

Work at the speed of your thoughts, not the speed of your fingers.
Open-source AI voice engine for Windows.
Local-first. Private. Model-agnostic.

Join the WaitlistWindows • Free Trial

macOSSoon™

iOSSoon™

AndroidSoon™

Three Inputs. Infinite Output.

Voice, Text Selection, and Vision are the pillars.

Input: VoiceCtrl+Alt+D

Dictate

Your voice becomes text, instantly, into any window. RAW Mode for unfiltered speed. Professional Mode for AI-polished output.

Input: Text SelectionCtrl+Alt+R

Refine

Select any text, speak your intent. Rewrite emails, fix grammar, translate, restructure. The text updates in place.

Input: ScreenCtrl+Alt+S

Vision

Capture any region of your screen — describe it, extract text, read tables, or ask questions about what you see.

AskCtrl+Alt+A

Query your local LLM for definitions, math, or quick facts without leaving your window.

NotesCtrl+Alt+N

Capture thoughts with automatic timestamps and metatags. Sequential formatting ready for batch processing.

ConnectorsSoon

Route output to Notion, Slack, webhooks, and more.

dIKta.me Versus The Cloud.

Why settle for rental when you can own the factory?

Feature comparison: dIKta.me vs cloud dictation services
Feature	The Others	dIKta.me
Privacy	Sent to servers	100% Local (Air-Gapped)
Speed	800-1200ms Latency/Inference	Sub-second Local Pipeline *
Consistency	Varies by Web Traffic	Hardware-Bound (Stable)
Censorship	Aggressive Filtering	Zero Restrictions
Cost	$96–180 / year	$20 (Once)
Word Limits	Capped	Unlimited while on Local
Offline Use	Requires WiFi	Works in a Bunker
Apps Needed	3–4 separate subscriptions	Just this one
* Whisper V3 Turbo + Gemma 3 on Vulkan GPU. Results vary by hardware and recording length.

Seen enough?

Try Free

The Specs.

Everything you need to make your voice heard. Nothing you don't.

Core

Whisper V3 Turbo

State-of-the-art local STT with natural tolerance for filler words, stutters, and punctuation.

Intelligence

Local Small Model Brain

Intelligent formatting, filler removal, and general knowledge. All local, all private, all fast.

Speed

Global Configurable Hotkeys

Ctrl+Alt+D (Dictate)

Ctrl+Alt+A (Ask)

Ctrl+Alt+R (Refine)

Ctrl+Alt+N (Notes)

More...

Integration

Precision Injection

Ultra-fast text injection with native window emulation across every application.

Workflow

Intelligent Modes

Standard - Direct Injection

Professional - Polished Output

Prompt - AI Instruction mode

Chat - Floating Overlay

Audio

Audio Ducking

Automatically reduces background volume from other applications during active recording sessions.

Power User

API Freedom

No markup, no middleman. Plug in your own OpenAI, Anthropic, or DeepSeek keys directly.

Translation

Cross-Lingual Engine

Real-time ES ↔ EN translation integrated at the model level for zero-latency bridging.

Privacy

Local Privacy Guard

100% local processing with automatic PII and sensitive data scrubbing from all logs.

Security

Encrypted SafeStorage

Secrets and keys are managed via OS-level AES-256 encryption. Never exposed, never raw.

Floating Pill

Minimal overlay that stays out of your way. Lives in the system tray when not recording.

Deploy

Self-Contained Binary

Lightweight self-contained .exe built with native C# and WinUI 3. No external runtime required.

Productivity

Voice Macros

Define trigger phrases that auto-expand into full macros, signatures, or templates.

Flexibility

Model Agnostic

Switch freely between Llama, Mistral, Gemma, or any Ollama-compatible model. No lock-in.

Reliability

Smart Fallback

Intelligent session recovery—never lose a word of dictation if a model hangs or crashes.

Voice

Text-to-Speech

Hear AI responses, read selections aloud, or listen to translations. Fully local via Kokoro ONNX.

Streaming

Real-Time STT

Live streaming speech-to-text via Deepgram for instant cloud transcription with sub-second latency.

Cloud

Pay-As-You-Go Wallet

No subscriptions. Load credits and use cloud STT + LLM at cost. $5 gets you ~65,000 words.

Setup

First-Run Wizard

Guided setup that installs Ollama, downloads models, and configures everything in minutes.

Performance

GPU Acceleration

Vulkan-powered Whisper inference for blazing local STT. Falls back to CPU gracefully when needed.

Customization

Custom Modes

Create your own dictation modes with custom system prompts, model overrides, and hotkey bindings.

Data

Session History

Every dictation logged locally in SQLite — word counts, latency, mode used. Your data, queryable.

Notifications

Spoken Alerts

AI responses and translations read aloud via system notifications. Hear answers without looking.

Transparency

Source Available

Inspect every line. No black boxes, no telemetry, no hidden network calls. Built in the open.

Ready to stop typing?

Join the Waitlist

NEW FEATURE

We Speak Spanish.
Hablamos Inglés.

Real-time translation at the edge. Dictate freely in your native tongue and let the engine bridge the gap instantly.

✓ No extra API keys required
✓ English ↔ Spanish (bidirectional)
✓ Context-aware translation

Input (Microphone)

Hola, necesito enviar el reporte financiero antes del mediodía, ¿puedes revisarlo?

Output (Application)

Hi, I need to send the financial report before noon, can you review it?

Input (Microphone)Ctrl+Alt+A

"What is the capital of madagascar and its population"

Output (Clipboard/Type)

"Antananarivo, 4,413,000 as of early 2026"

JUST ADDED

Don't Search. Just Ask.

Skip the browser tab. Query your local LLM instantly for definitions, math, or quick facts without ever leaving your flow.

✓ Instant Q&A - Math, quotes, refs
✓ No Tracking - Local execution
✓ Zero Context Switch - Stay in flow

QUICK CHAT

Your AI, Always On Call.

A floating chat overlay powered by any model — local or cloud. Rich Markdown responses, conversation history, and zero context switching.

✓ Floating Overlay — Stays above every app
✓ Model Agnostic — GPT, Claude, Gemini, Ollama
✓ Rich Output — Markdown, code blocks, links

Quick Chat

—□✕

gpt-5.2-pro▾

◷⚙⊕⊡+

📎

Type a message...

▶

Provider Settings

Active Model

▼

Anthropic Key

sk-ant-••••••••••••••••••

Google AI Key

AIzaSy••••••••••••••••••

"Encryption enabled. Keys never leave this machine."

YOUR KEYS

Your Tokens. Your Choice.

Use local, use API Keys, its your choice, not theirs. Anthropic, Google, Deepseek, or local Ollama — dIKta.me unifies them all.

✓ BYO Keys - No markup, ever.
✓ Local First - Ollama & Llama.cpp
✓ Infinite Flexibility - Switch in seconds

PRODUCTIVITY

Say Less. Type More.

Define voice-triggered macros that expand into full text blocks instantly. Signatures, templates, addresses — speak the trigger, get the output.

✓ Voice Triggers — "Inject [name]" pattern
✓ Instant Expansion — Multi-line output in milliseconds
✓ Unlimited Macros — Create as many as you need

Voice Macros

● Voice Input

↓ ↓ expands to

● Injected Text

"2 macros configured"

Kokoro TTS Engine

● Audio Output

● Speaking

The quick brown fox jumps over the lazy dog near the riverbank

Progress0%

Voice: af_heartKokoro ONNX · Local

VOICE OUTPUT

AI That Speaks Back.

Hear AI responses read aloud, listen to translations, or have any selected text spoken naturally — all processed locally.

✓ 100% Local — Kokoro ONNX, no cloud
✓ Natural Voices — Multiple voice presets
✓ Read Anything — Selections, responses, translations

That was just the highlight reel.

See Pricing

Use it with...

TERMINAL
CURSOR
ANTIGRAVITY
GOOGLE DOCS
WHATSAPP
NOTEPAD
VS CODE
SLACK
EXCEL
DISCORD
OUTLOOK
CHROME
ARC

Common Questions

The short answers. No fluff.

Stop Renting Software.

No monthly fees. No word limits. Just great tools you own.

Free Trial

Try all 8 modes right now. No card required.

✓ All 8 workflow modes included
✓ Cloud STT & Cloud LLM
✓ Quick Chat & Vision included
✓ Custom Presets & Macros
✓ Cloud credits included to get started
✕ Upgrade anytime for local AI + offline

Try Free

Cloud credits included. Top up anytime.

Launch Price

Full Version

$25$20

The full app. On your hardware. Yours forever.

✓ Local AI (Whisper + Ollama + Kokoro)
✓ BYOK — bring your own API keys
✓ Offline mode — works without internet
✓ All modes + all features unlocked
✓ Sub-second local pipeline (STT + LLM)
✓ MIT Open Source License

Get Full Version — $20

One-time payment. Yours forever.

Build from Source

MIT licensed. Clone, build, run. No license needed for self-compiled builds.

📚 Full Source Code on GitHub
📖 Build Guide in CONTRIBUTING.md

View on GitHub

Support the DeveloperKeep dIKta.me updated & independent via Ko-Fi

STOP TYPING.STARTTALKING.THINKING.WORKING.WINNING.

Three Inputs. Infinite Output.

Dictate

Refine

Vision

dIKta.me Versus The Cloud.

The Specs.

Whisper V3 Turbo

Local Small Model Brain

Global Configurable Hotkeys

Precision Injection

Intelligent Modes

Audio Ducking

API Freedom

Cross-Lingual Engine

Local Privacy Guard

Encrypted SafeStorage

Floating Pill

Self-Contained Binary

Voice Macros

Model Agnostic

Smart Fallback

Text-to-Speech

Real-Time STT

Pay-As-You-Go Wallet

First-Run Wizard

GPU Acceleration

Custom Modes

Session History

Spoken Alerts

Source Available

We Speak Spanish.Hablamos Inglés.

Don't Search. Just Ask.

Your AI, Always On Call.

Your Tokens. Your Choice.

Say Less. Type More.

AI That Speaks Back.

Common Questions

Stop Renting Software.

Free Trial

Full Version

Build from Source

STOP TYPING.
START
TALKING.
THINKING.
WORKING.
WINNING.

We Speak Spanish.
Hablamos Inglés.