Skip to main content

Use Cases Overview

hns transcribes voice to text and writes the output to stdout. As a standard CLI utility, it composes seamlessly with pipes, redirects, and command substitutions, making it a universal voice input layer for any text-based workflow.

The examples below demonstrate core patterns. Once you understand these, you'll see hns can drive any CLI tool that accepts text input.

Turn Your Voice Into Polished Text​

Spoken language is messy. Pipe your raw hns transcription through a local or remote LLM to instantly fix grammar, remove filler words, and produce clean, professional text. Perfect for writing emails, Slack messages, or documentation without the friction of typing.

Quick Start:

# Clean up transcription using a remote LLM
llm --system "Fix grammar. Output only corrected text." "$(hns)"

Dive deeper in the full guide →

Drive AI Coding Agents​

Provide detailed, context-rich prompts to coding agents like Claude Code or cursor-agent without typing. Speaking your instructions is faster and encourages more comprehensive prompts, leading to better first-time responses from the AI.

Quick Start:

# Pipe your voice directly to Claude Code
claude "$(hns)"

See advanced workflows in the full guide →

Effortless Daily Journaling​

Overcome the "blank page" problem by speaking your thoughts freely. A simple command can append your raw transcription to a daily note. For more polished entries, you can pipe it through an LLM to structure your thoughts automatically.

Quick Start:

# Append raw transcription to a dated journal file
hns >> "$JOURNAL_DIR/$(date '+%Y-%m-%d').md"

Build a complete journaling system in the full guide →

The --last Flag

All workflows support the --last flag (e.g., hns --last). Use it to reprocess your previous recording without speaking again—perfect for when a command fails or you want to try a different model.

Beyond These Examples​

The use cases above are starting points. Since hns outputs plain text, it integrates with virtually any CLI workflow:

  • Git commits: git commit -m "$(hns)" for dictated commit messages
  • Translation: Integrate with llm, ollama, or llama-cli to speak in one language and output another
  • Email drafting: Integrate with mail to compose messages by voice
  • Any text-accepting CLI tool: If it reads stdin or arguments, hns can provide the input

The pattern is universal: speak your input, let hns transcribe it, and compose it with standard Unix tools.