Dictation Features
Explore the full range of Wispr Flow's dictation capabilities, from natural speech recognition and auto-punctuation to code dictation and context-aware formatting.
Natural Speech Recognition
Wispr Flow's AI engine is built to understand natural, conversational speech. Unlike traditional dictation systems that expect you to speak in a rigid, mechanical way, Wispr Flow handles:
- Filler words: "Um," "uh," "like," and other verbal fillers are automatically filtered out of the transcription
- Self-corrections: If you say something and then immediately correct yourself, the AI captures your intended meaning
- Natural pauses: Brief pauses do not break the flow — the AI knows when you are thinking vs. when you have finished a sentence
- Varied speaking speeds: Whether you speak quickly or slowly, accuracy remains consistent
What you actually say: "So um I think we should uh we need to update the landing page with the new uh the new pricing information before the launch on Friday" What Wispr Flow produces: I think we need to update the landing page with the new pricing information before the launch on Friday.
Auto-Punctuation and Formatting
One of Wispr Flow's most powerful features is intelligent auto-formatting. The AI analyzes your speech patterns to automatically insert:
| Formatting | How It Works |
|---|---|
| Periods | Inserted at natural sentence endings based on intonation and context |
| Commas | Added at clause boundaries, lists, and natural pause points |
| Question marks | Detected from rising intonation and question word patterns |
| Capitalization | Sentence beginnings, proper nouns, and acronyms capitalized automatically |
| Paragraph breaks | Inserted when you pause between distinct topics or ideas |
| Numbers | Context-aware: "three hundred" becomes 300 in data contexts, "three" in prose |
Code Dictation
Wispr Flow recognizes when you are in a code editor and switches to code-aware dictation mode. This is a game-changer for developers who want to reduce keyboard strain:
What you say: "define a function called calculate total that takes a list of prices and a tax rate parameter with default value zero point zero eight" What appears in VS Code: def calculate_total(prices: list, tax_rate: float = 0.08): pass
Code dictation features include:
- Language detection: Automatically detects Python, JavaScript, TypeScript, Java, and other languages from your file context
- Naming conventions: Applies snake_case, camelCase, or PascalCase based on the language and context
- Symbol handling: Understands "equals," "open parenthesis," "bracket," "arrow function," and other code vocabulary
- Indentation: Maintains proper indentation levels based on your code structure
Email Composition
When Wispr Flow detects you are in an email client (Gmail, Outlook, Apple Mail), it adjusts its formatting to produce well-structured emails:
What you say: "hi team just a quick update the client approved the design mockups we can move forward with development starting Monday please review the specs document I shared yesterday and let me know if you have any questions thanks" What appears in Gmail: Hi team, Just a quick update - the client approved the design mockups! We can move forward with development starting Monday. Please review the specs document I shared yesterday and let me know if you have any questions. Thanks!
Document Writing
For longer-form content in Google Docs, Notion, Word, or other writing tools, Wispr Flow structures your dictation into well-organized prose with proper headings, paragraphs, and formatting.
- Paragraph detection: Topic shifts are automatically separated into new paragraphs
- List creation: When you enumerate items ("first... second... third..."), Wispr Flow can format them as a list
- Heading detection: Say "heading" or "title" followed by text to create formatted headings
- Quote formatting: Phrases like "quote" or "in their words" can trigger block quote formatting
Multi-Language Support
Wispr Flow supports dictation in multiple languages. The AI automatically detects which language you are speaking and transcribes accordingly:
- English (US, UK, Australian, and other variants)
- Spanish, French, German, Italian, Portuguese
- Japanese, Korean, Chinese (Mandarin)
- And many more — with new languages being added regularly
Context-Aware Formatting
What makes Wispr Flow truly stand out is its awareness of where you are typing. The same spoken input can produce different formatted output depending on the application:
| Application | Formatting Behavior |
|---|---|
| VS Code | Code syntax, proper indentation, variable naming conventions |
| Gmail | Email structure, greeting/closing formatting, professional tone |
| Slack / Discord | Casual tone, shorter sentences, emoji-friendly |
| Google Docs | Long-form prose, paragraph structure, heading detection |
| Terminal | Command syntax, flags, and arguments formatted correctly |
| Browser URL bar | No spaces, treats input as search query or URL |
Correction Commands
Made a mistake? Wispr Flow provides several ways to correct dictated text:
- "Undo that" — Removes the last dictation
- "Replace [word] with [word]" — Swaps a specific word
- "Delete last sentence" — Removes the last complete sentence
- "Scratch that" — Clears the current dictation before it is inserted
- "Correct [word]" — Shows alternative transcriptions for a specific word
💡 Try It: Context-Aware Dictation
Open three different applications — a code editor, an email client, and a messaging app. Dictate the same general idea in each and compare how Wispr Flow formats the output differently. Notice the tone, punctuation, and structure changes.