speech-to-text

name	speech-to-text
description	Open-source speech-to-text and voice typing tools for dictation, transcription, and voice input across desktop, mobile, and terminal. USE FOR: evaluating local/hybrid ASR tools, comparing platform-specific voice typing options, selecting dictation tools with direct editor integration. DO NOT USE FOR: meeting bots, call transcription services, closed-source dictation products, general speech APIs without a typing interface, non-open-source mobile assistants. You MUST load this skill when evaluating or comparing open-source speech-to-text and voice typing tools.
license	MIT

Speech to Text

WHEN TO USE

• You need open-source tools for dictation, voice typing, or speech-to-text text entry while actively working.
• You are comparing desktop, mobile, or terminal-first speech input workflows.
• You want tools that can insert recognized text directly into an editor, input field, or command-line workflow.
• You are specifically evaluating local-only or hybrid speech recognition options.
• You need a platform-specific shortlist for Linux, macOS, Windows, Android, or iOS voice typing tools.

WHEN NOT TO USE

• You are looking for closed-source dictation or transcription products.
• You need meeting bots, note takers, or call transcription services rather than active voice typing tools.
• You want general speech-to-text APIs or backend transcription services without a usable typing interface.
• You are evaluating non-open-source mobile assistants or OS-native features that are not open-source projects.
• You need a comprehensive survey of all transcription software rather than tools focused on dictation workflows.

Scope

This list focuses on tools that let you speak and get text into the place where you are actively working.

• Open-source apps, keyboards, menu bar utilities, and CLI tools for dictation or voice typing
• Desktop, mobile, and terminal-first workflows
• Local-only and hybrid tools, as long as the project itself is open source

This list does not try to cover:

• Closed-source products
• General transcription tools that do not support typing or dictation workflows
• Meeting bots, note takers, or speech APIs without a usable typing interface

Core Process

1. Identify Target Platform: Determine the OS (Linux, macOS, Windows, Android, iOS).
2. Select Mode: Choose Local for fully offline recognition or Hybrid for cloud-assisted ASR.
3. Filter by Engine: Match against available ASR engines (Whisper, Whisper.cpp, Faster Whisper, Parakeet, Vosk).
4. Evaluate Tool: Scan the directory below for matching entries, then visit the repository for setup instructions.

Quick Start Example

# Transcribe an audio file using whisper.cpp
git clone https://github.com/ggml-org/whisper.cpp
cd whisper.cpp && make -j
./main -m models/ggml-base.en.bin -f sample.wav

# Start nerd-dictation on Linux (offline Vosk-based dictation)
nerd-dictation begin --vosk-model en-us-small
# Speak into your microphone...
nerd-dictation end
# Transcribed text appears in the active window.

Most tools on this list support offline speech recognition. See Mode and Engine in the directory below for details.

Directory

Mode: Local means on-device speech recognition. Hybrid means the tool supports both local and cloud or BYO-cloud setups.

Name	Platforms	Mode	Engine	Summary
Amical	macOS, Windows	Local	Whisper	Context-aware dictation that adapts formatting to the app you are using.
Epicenter Whispering	Linux, macOS, Windows, Web	Hybrid	Whisper	Local-first dictation with global shortcut and multiple Whisper providers.
FluidVoice	macOS	Hybrid	Parakeet, Apple Speech, Whisper	macOS dictation that types into any app and switches between local speech engines.
FnKey	macOS	Hybrid	Deepgram Nova-3, Groq Whisper	Rust menu bar app; activates mic only while holding Fn, real-time streaming.
hyprwhspr	Linux	Hybrid	Whisper.cpp, Parakeet	Push-to-talk Linux dictation with visualizer, Waybar, and systemd integration.
nerd-dictation	Linux	Local	Vosk	Offline dictation that types into any window via simulated keystrokes.
OmniDictate	Windows	Local	Whisper	Desktop dictation tool aimed at type-anywhere workflows.
Tambourine Voice	macOS, Windows	Hybrid	Faster Whisper	Voice interface for any app with configurable STT and LLM providers.
Vibe	Linux, macOS, Windows	Local	Whisper.cpp	Desktop app to transcribe audio and video offline.
VoiceInk	macOS	Hybrid	WhisperKit	Native macOS dictation with per-app tuning and custom dictionary.
VoiceTypr	macOS, Windows

Related Projects

Speech recognition engines, models, and APIs that power the tools listed above.

• whisper.cpp - Pure C/C++ Whisper inference; runs on Apple Silicon, CUDA, Vulkan, and WASM.
• Faster Whisper - CTranslate2-based Whisper reimplementation; up to 4x faster with lower memory.
• Apple Speech - Apple's on-device speech recognition framework for iOS, macOS, and watchOS.

References