Voice Typing vs Traditional Typing: A Productivity Experiment

The Experiment

Everyone claims voice typing is faster. But how much faster? And is faster always better?

We set up a controlled experiment: five real-world tasks, each completed twice — once by typing on a keyboard and once by voice typing with Murmur. Same person, same content, same quality standards.

Here is exactly what happened.

The Setup

Participant: A software developer who types at approximately 65 WPM (above average) and has two weeks of experience with voice typing.

Voice typing tool: Murmur with Ctrl+Space activation.

Tasks selected: Representing the five most common text input scenarios for knowledge workers.

Rules:

• Each task is completed to the same quality standard
• Time includes any corrections or edits needed
• The final output must be equivalent regardless of input method
• Each task was done twice with each method, and the average time is reported

Task 1: Email (Professional, 150 words)

Scenario: Reply to a client asking about a project delay.

Typing

• Time: 4 minutes 12 seconds
• Process: Started typing, paused to think about wording, backspaced to rephrase the tone around the delay, added closing.
• Edits: 3 mid-stream corrections

Voice Typing

• Time: 1 minute 48 seconds
• Process: Spoke the email in one take, reviewed for 20 seconds, fixed one word.
• Edits: 1 post-dictation correction

Result

Why voice won: Email is conversational content. Speaking it produced a natural, appropriate tone with minimal editing. The AI-powered transcription produced clean output with proper greeting and sign-off structure.

---

Task 2: Code Comment (Technical, 80 words)

Scenario: Write a documentation comment for a function that handles user authentication with JWT tokens.

Typing

• Time: 1 minute 35 seconds
• Process: Typed the comment block, including parameter descriptions and return value documentation. Used code-specific formatting naturally.
• Edits: 0 corrections needed

Voice Typing

• Time: 1 minute 22 seconds
• Process: Dictated the comment. Had to spell out "JWT" and correct "parameter" which was transcribed as "perimeter" on first attempt.
• Edits: 2 post-dictation corrections

Result

Analysis: Voice was marginally faster, but the corrections ate into the time advantage. For short, technical content with specialized vocabulary, the speed difference is minimal. Typing might be preferable for code comments specifically because of the formatting requirements (backticks, parameter names, etc.).

---

Task 3: Slack Messages (Casual, 5 messages totaling 120 words)

Scenario: Reply to five different Slack threads — a standup update, a code review response, a meeting logistics message, a casual team chat, and a quick question.

Typing

• Time: 3 minutes 45 seconds
• Process: Switched between threads, typed each reply. Short messages but frequent context switching.
• Edits: 1 typo correction

Voice Typing

• Time: 1 minute 30 seconds
• Process: Navigated to each thread, pressed Ctrl+Space, spoke the reply, moved on. The casual tone matched exactly what you would want in Slack.
• Edits: 0 corrections needed

Result

Why voice dominated: Short, casual messages are the ideal voice typing use case. No complex vocabulary, conversational tone matches natural speech, and the sheer number of messages amplifies the per-message time savings.

---

Task 4: Documentation (Technical/formal, 400 words)

Scenario: Write a README section explaining how to set up a development environment.

Typing

• Time: 12 minutes 20 seconds
• Process: Typed the documentation, including code blocks (terminal commands), numbered steps, and configuration examples. Frequently paused to check accuracy of commands.
• Edits: Multiple mid-stream restructures

Voice Typing

• Time: 7 minutes 45 seconds
• Process: Dictated the prose sections (setup instructions, explanations), then typed the code blocks and terminal commands manually. Hybrid approach.
• Edits: 3 post-dictation corrections, plus manual code block formatting

Result

Key insight: Pure voice typing does not work well for documentation that includes code blocks and terminal commands. The hybrid approach — dictate the prose, type the code — captured most of the speed advantage while maintaining formatting accuracy.

---

Task 5: AI Prompt (Conversational/technical, 200 words)

Scenario: Write a detailed prompt for Claude Code asking it to refactor an authentication module.

Typing

• Time: 5 minutes 10 seconds
• Process: Typed the prompt, which required describing the current codebase structure, the desired changes, and constraints. Paused several times to organize thoughts.
• Edits: 2 restructures

Voice Typing

• Time: 2 minutes 40 seconds
• Process: Spoke the prompt conversationally, explaining what the AI should do as if briefing a colleague. The natural speaking style produced a clear, well-structured prompt.
• Edits: 1 minor correction

Result

Why voice excelled: AI prompts are essentially instructions, and we give instructions by speaking more naturally than by typing. The voice-typed prompt was actually more detailed and clearer because speaking allowed for a more natural explanatory flow.

---

Aggregate Results

Overall: voice typing was 44% faster across all tasks combined.

When Voice Typing Wins

Voice typing has a clear advantage when:

• The content is conversational. Emails, Slack messages, AI prompts, and explanatory text flow naturally from speech.
• The text is more than a sentence. Very short inputs (a search query, a filename) are not worth activating voice typing for.
• Tone matters. Paradoxically, voice produces better tone for emails and messages because you naturally modulate your communication style when speaking.
• You need to explain something. Explanations, instructions, and descriptions come out more clearly when spoken.

When Keyboard Wins

Stick with typing when:

• The content is heavily formatted. Code, markdown tables, mathematical notation, and structured data are faster to type.
• You need precise technical terms. Variable names, CLI flags, and abbreviations are more reliably typed.
• The input is very short. Under 10 words, typing is faster due to voice typing activation overhead.
• You are in a noisy or public environment. Practical limitation, not a productivity one.

The Hybrid Approach: Best of Both Worlds

The experiment's most useful finding was not that voice is always faster — it is that the combination of voice and keyboard is faster than either alone.

The optimal workflow looks like this:

1. Voice for anything conversational, explanatory, or more than a sentence
2. Keyboard for code, formatting, short inputs, and precise technical content
3. Voice to draft, then keyboard to refine

This hybrid approach captured an average speed improvement of 44% across our five tasks — and that number would likely increase with more voice typing experience.

Try It Yourself

The best way to evaluate voice typing is to run your own experiment. Here is a simple test:

1. Pick three tasks you do daily (email, Slack, documentation, etc.)
2. Time yourself completing each by typing
3. Download Murmur and complete the same tasks by voice
4. Compare the results

Most people find the speed difference convincing after a single test. After a week of practice, the gap widens further.

---

Want to run your own productivity experiment? Get Murmur free and see how much faster you can work.