Have the text inserted right by your cursor when using the shortcut.
Start and stop recording via shortcut and DirektText inserts the result right where you were working.
DirektText helps you record speech, choose the output format, and insert the result right where you work. Dictation, transcription, modes, history, vocabulary, Quick Phrases, file import, and local or ready-to-use cloud processing live in one Windows app.
The same rough spoken input can become a clean paragraph, meeting recap, email draft, SOP draft, structured note, transcript, or custom output.
Subscriptions include ready-to-use cloud transcription capacity. You can also use supported local models on your machine or BYOK.
Ready-to-use cloud transcription and processing capacity included with every subscription.
Local transcription via Whisper Turbo v3 or Parakeet and supports local LLMs via Ollama or LM Studio.
Bring your own keys for supported ASR & LLM providers (OpenAI, AssemblyAI, ElevenLabs, Deepgram, Mistral, Google, Anthropic).
Start and stop recording via shortcut and DirektText inserts the result right where you were working.
Record with a visible control, see the output in the overlay, then copy or close it without automatically pasting.
Capture your own dictation, notes, explanations, and voice input. This is the most common case for everyday work.
Capture audio from calls, recordings, or other on-screen sources where appropriate.
Capture both sides of a conversation when you need the full context.
Meeting and system audio should be captured with consent.
Names, acronyms, products, and work-specific terms are remembered so DirektText can better understand the words you actually use.
Set up automatic find and replace to turn trigger words into phrases and messages you often send.
Format common spoken forms automatically, then add more specific regex rules for names, phrases, punctuation, numbers, or recurring wording patterns.
\b(twenty five|25) percent\b -> 25%
Find previous recordings, copy the text again, or change the format by reprocessing it with a different mode. You can also name recordings to make them easier to find.
Includes 1 hour of fast transcription and 30 minutes of premium transcription.