Golosok
1 like
Native 100% offline macOS voice dictation & speech-to-text (Apple Silicon). Powered by Sber GigaAM v3 & Whisper. Instant direct input under cursor via ?+Space, Hold-to-talk, zero cloud. Free Wispr Flow & Superwhisper alternative.
Cost / License
- Free
- Open Source (MIT)
Platforms
- Mac
Features
- Speech Recognition
- Swift App
- Speech to text
- Dictation
Golosok News & Activities
Highlights All activities
Recent activities
subtiliorars-sys added Golosok as alternative to Privacy Scribe- Rostislav-7208 updated Golosok
- Rostislav-7208 added Golosok
- Rostislav-7208 liked Golosok
Rostislav-7208 added Golosok as alternative to Superwhisper, Vibe Transcribe, OpenWhispr and TypeWhisper
Golosok information
Developed by
Berliner187Licensing
Open Source (MIT) and Free product.Written in
Alternatives
13 alternatives listedSupported Languages
- English
- Russian
AlternativeTo Category
AI Tools & ServicesGitHub repository
- 2 Stars
- 0 Forks
- 2 Open Issues
- Updated
No comments or reviews, maybe you want to be first?
What is Golosok?
Golosok is an open-source, privacy-first macOS utility engineered for instant voice-to-text dictation and batch media transcription. It runs 100% locally on-device using CoreML and the Apple Neural Engine (ANE). Zero cloud processing, zero subscription fees, zero privacy risks.
Key Features
- Direct Input Under Cursor: Press ? + Space in any active application (VS Code, Notion, Telegram, Xcode, Obsidian, Slack) — dictate — text drops instantly at your active cursor position.
- Dual Dictation Modes:
- Toggle Mode: Press hotkey once to start, press again to finish and paste.
- Push-to-Talk / Hold Mode: Hold hotkey while speaking, release to instantly transcribe and paste. Zero Cloud & Total Privacy: Audio streams and transcribed text never leave your machine. Ideal for confidential notes and NDA meeting records.
- Hardware Accelerated: Optimized for Apple Silicon (M1–M5). Inference is offloaded to the Apple Neural Engine (ANE) via CoreML, preserving CPU/GPU cycles and battery life.
- Multi-Engine Support: Integrates Sber GigaAM v3 for state-of-the-art Russian/Runglish/IT-slang accuracy and Whisper Large v3 / Medium for English and multilingual dictation.
- Batch Media Transcription: Drag-and-drop offline processing for audio and video files (MP3, WAV, M4A, MP4, WebM).