AI Tools
Audio AIIntermediate

VoiceInk

VoiceInk is an on-device voice-to-text and automatic text conversion solution exclusively for macOS, developed by Beingpax (Pax) and released on March 15, 2026.

VoiceInk is an on-device voice-to-text and automated text conversion solution for macOS, developed by Beingpax (Pax) and released on March 15, 2026. This tool utilizes OpenAI's Whisper speech recognition model, enhanced and ported to C/C++ as whisper.cpp, along with the FluidAudio library as its engine. It performs real-time, high-speed transcription processing within the macOS device without network transmission. Similar to recording conversations directly onto paper in an underground bunker with no internet access, VoiceInk adopts a local-first architecture, encapsulating the user's audio signals within local hardware (CPU/GPU) and performing computations without transmitting data to cloud servers.

Existing cloud-based services transmit all input voice data to external servers in real-time for storage and processing, constantly posing the risk of sensitive, private data or personal information being exposed due to hacking or misuse, and also incurring significant monthly subscription fees. VoiceInk is a powerful open-source, native macOS application that completely resolves these privacy concerns and ongoing cost barriers. In particular, by introducing a Power Mode feature that automatically adjusts appropriate conversion rules and prompt contexts based on the user's current active application or specific website, it goes beyond a simple text converter and acts as a smart writing assistant, intelligently recognizing the user's screen situation and refining sentences into optimal words and formats.

In biotechnology and medical research, researchers require a hands-free recording environment to prevent contamination during experiments, as manipulating mice and keyboards can be difficult. When VoiceInk is running, researchers can record observation values and cell states in real-time offline using only their voice without interrupting their work in the clean bench or microscope observation. This data is immediately saved as a text Markdown document on the local drive. In particular, unique research terms (e.g., 'CRISPR-Cas9', 'Drosophila melanogaster'), names of individuals, and development code names, which are often mispronounced or have recognition errors, can be registered in a personal vocabulary dictionary to drastically reduce misrecognition rates. When combined with external API integration, it can implement an all-in-one pipeline for paper translation and draft summarization.

💻 System Requirements

🧠RAM

0 (Operates using Apple Silicon Unified Memory shared architecture)

💾Storage

Approximately 150MB for the app package and basic libraries; additional storage required depending on the size of the Whisper model file (minimum 75MB, maximum 3.1GB)

Installation

4-1. Quick Start

brew install --cask voiceink

4-2. Detailed Installation

Clone the GitHub repository

git clone https://github.com/Beingpax/VoiceInk.git cd VoiceInk

Open the project in Xcode to build the Swift project

open VoiceInk.xcodeproj

After completing the local compilation by running Product > Build (Cmd + B) from the Xcode top menu, execute the application.

🧬 Bio Use Cases

🔬

Hands-free research logging within animal labs and clean benches

During cell culture and reagent mixing experiments, use voice commands with gloves on to record observations (e.g., cell density 85%, viability 98%) in real-time and save them as a local text markdown file.

🧬

De-identified transcription and data protection for clinical patient interviews

In a completely isolated hospital laboratory with no network connection, real-time transcription of patient speech data is performed using Apple Silicon GPU acceleration, and de-identified clinical notes are immediately created in a local database without any data being sent to external servers.

💊

Real-time translation and summarization for multilingual research seminars

Translate audio data from offline seminars lasting 10 minutes or longer with researchers from overseas subsidiaries using the Whisper multilingual model, and then connect it to a local AI model API to automatically convert key strategic decision-making items into a summary of 30 seconds or less.

FAQ

What is VoiceInk?

VoiceInk is an on-device voice-to-text and automated text conversion solution for macOS, developed by Beingpax (Pax) and released on March 15, 2026. This tool utilizes OpenAI's Whisper speech recognition model, enhanced and ported to C/C++ as whisper.cpp, along with the FluidAudio library as its engine. It performs real-time, high-speed transcription processing within the macOS device without network transmission. Similar to recording conversations directly onto paper in an underground bunker with no internet access, VoiceInk adopts a local-first architecture, encapsulating the user's audio signals within local hardware (CPU/GPU) and performing computations without transmitting data to cloud servers. Existing cloud-based services transmit all input voice data to external servers in real-time for storage and processing, constantly posing the risk of sensitive, private data or personal information being exposed due to hacking or misuse, and also incurring significant monthly subscription fees. VoiceInk is a powerful open-source, native macOS application that completely resolves these privacy concerns and ongoing cost barriers. In particular, by introducing a Power Mode feature that automatically adjusts appropriate conversion rules and prompt contexts based on the user's current active application or specific website, it goes beyond a simple text converter and acts as a smart writing assistant, intelligently recognizing the user's screen situation and refining sentences into optimal words and formats. In biotechnology and medical research, researchers require a hands-free recording environment to prevent contamination during experiments, as manipulating mice and keyboards can be difficult. When VoiceInk is running, researchers can record observation values and cell states in real-time offline using only their voice without interrupting their work in the clean bench or microscope observation. This data is immediately saved as a text Markdown document on the local drive. In particular, unique research terms (e.g., 'CRISPR-Cas9', 'Drosophila melanogaster'), names of individuals, and development code names, which are often mispronounced or have recognition errors, can be registered in a personal vocabulary dictionary to drastically reduce misrecognition rates. When combined with external API integration, it can implement an all-in-one pipeline for paper translation and draft summarization.

When should I use VoiceInk?

VoiceInk is an on-device voice-to-text and automatic text conversion solution exclusively for macOS, developed by Beingpax (Pax) and released on March 15, 2026.

What is a biomedical use case for VoiceInk?

Hands-free research logging within animal labs and clean benches: During cell culture and reagent mixing experiments, use voice commands with gloves on to record observations (e.g., cell density 85%, viability 98%) in real-time and save them as a local text markdown file.

📄 Official Docs🐙 GitHub

📝 Update Notes

  1. vv2.07/16/2026

    VoiceInk 2.0은 개인화된 워크플로우를 위한 '모드' 기능과 더 다양한 AI 모델 지원을 통해 한층 강력해졌습니다. 특히 확장된 모델 선택권을 통해 생명공학 연구의 핵심인 복잡한 전문 용어와 실험 데이터를 더욱 정교하게 텍스트로 변환할 수 있습니다. 새롭게 도입된 어시스턴트 기능으로 기록된 내용에 대해 대화하며 맥락 있는 후속 질문이 가능해져, 실험 노트 정리와 데이터 분석의 효율을 높여줍니다. 또한 실시간 기록의 안정성이 개선되어 실험 중 발생하는 중요한 관찰 사항을 놓치지 않고 안전하게 기록할 수 있습니다.

  2. vv1.797/10/2026

    VoiceInk v1.79에서는 AppleScript를 이용한 붙여넣기 과정에서 발생하던 갑작스러운 충돌 문제가 해결되었습니다. 또한, Fn 단축키가 의도치 않게 작동하던 오류도 수정되어 더욱 안정적인 사용이 가능합니다. 실험 데이터를 자동화된 스크립트로 처리하거나 음성 기록을 남길 때, 흐름이 끊기지 않고 정확하게 작업할 수 있어 연구 효율을 높이는 데 도움이 될 것입니다.

🧪 Related Code of Life

No related Code of Life posts yet.