--- title: "VoiceMe — On-Device AI Subtitles" description: "Runs Whisper speech recognition locally, turning live microphone audio into text in real time. No audio ever leaves your machine. Windows and macOS." url: "https://royalsoftworks.com/products/voiceme/" source: "https://royalsoftworks.com" format: "markdown" note: "Markdown rendering of the HTML page at `url`. Same content, same canonical URL." --- [All products](https://royalsoftworks.com/#products) Speech Recognition Coming Soon # VoiceMe Real-time subtitles. No audio leaves your machine. On-device Whisper ASR turns live audio into live subtitles, overlaid on any application — meetings, calls, screen recordings. Three model sizes for any hardware. Zero audio leaves your machine, ever. [Join the waitlist](https://royalsoftworks.com/contact/) [View pricing](https://royalsoftworks.com/pricing/) On-device Whisper inference 3 Model sizes Live Subtitle stream 0 Audio transmitted WHISPER ON-DEVICE REAL-TIME SUBTITLES TINY BASE SMALL NO AUDIO UPLOAD WIN + MAC SENSITIVE MEETING SAFE LIVE STREAM GPU ACCELERATED WHISPER ON-DEVICE REAL-TIME SUBTITLES TINY BASE SMALL NO AUDIO UPLOAD WIN + MAC SENSITIVE MEETING SAFE LIVE STREAM GPU ACCELERATED Available on - Windows 10 / 11 / Server - macOS 12 Monterey+ Core capabilities ## What it does. 01 Local Whisper ### Speech recognition that never touches the cloud. OpenAI's Whisper architecture, running on your CPU or GPU. Three model sizes — Tiny for ambient transcription, Base for general use, Small for accuracy-critical work. Multi-language support, automatic language detection. The kind of tool you can run during legally privileged or commercially sensitive meetings without having to ask anyone's permission. - Whisper Tiny, Base, Small models bundled - CPU and GPU inference paths (CUDA, Metal, ROCm) - 99 languages with automatic detection - Streaming inference — words appear as spoken 01 02 Always-on overlay ### Subtitles float over whatever you're doing. A non-blocking overlay window stays on top of any application — Zoom, Teams, browser tabs, documents. Move, resize, recolour and reposition without interrupting the underlying app. Configurable hide-on-idle so the subtitles disappear gracefully when no one's talking. - Always-on-top overlay across any app - Per-application positioning memory - Customisable font, size, colour, opacity - Hide on idle, reappear on speech - Global hot-keys for start / pause / hide / clear 02 Document Live subtitles Document Meeting mode Document Hotkey: ⌘⇧V 03 Export transcripts ### Save what was said. As plain text or SRT. Optional transcript saving. Plain text for notes, SRT for video subtitling, JSON for downstream pipelines. Speaker diarisation (separating distinct voices) on the Small model. Useful for compliance teams who need a record without paying a vendor per minute of audio. - Plain text, SRT, and JSON export - Speaker diarisation on the Small model - Per-segment timestamps preserved - Optional automatic save to chosen folder - Off by default — audio is processed in memory unless you opt in 03 assistant $ voiceme export --format srt ▸ Detecting speakers... ▸ Diarising: 3 voices found. ▸ Writing meeting\_2026-05-19.srt ▸ Done. 412 segments. More capabilities ## Everything else it does. ### Privacy by default No audio is recorded unless you explicitly enable transcript saving. Even then, files stay on your machine. ### Hot-key control Bind start, pause, hide and clear to global hot-keys. Useful for impromptu transcription during calls. ### Multi-language Whisper's full language coverage — 99 languages with varying accuracy. Use automatic detection or pin a specific language. ### GPU-aware Detects available GPU (Apple Silicon, NVIDIA CUDA, ROCm). Falls back gracefully to CPU on machines without a compatible GPU. Also in the suite [### AssistantGeneral Your organisation's knowledge, at every employee's fingertips. View product](https://royalsoftworks.com/products/assistant-general/) [### OfficeBrief Find any internal document in under a second. View product](https://royalsoftworks.com/products/office-brief/) [### Privatta Send sensitive files without routing through someone else's cloud. View product](https://royalsoftworks.com/products/privatta/) Frequently asked ## Questions we hear often. 01 Does VoiceMe send any audio to a server? No. All speech recognition runs on your machine via local Whisper inference. Audio is processed in memory and discarded unless you explicitly enable transcript saving. 02 Which Whisper model should I use? Tiny: fastest, OK for ambient note-taking. Base: balanced, good for most general use. Small: highest accuracy, recommended for meeting transcription and any case where words must be right. 03 What languages does it support? All 99 languages Whisper supports, with varying accuracy. Use automatic detection or pin a specific language for better results. 04 When will it be available? VoiceMe is in development. Join the waitlist via the contact form to be notified when general availability is announced. ## Talk to the team that actually builds the software. Pilots, licensing, demos, security questionnaires, or a question you are not sure is a question yet. All of it lands with engineers and product leads rather than a routing layer, and none of it starts a drip campaign. Rather not talk to anyone?[Download it free](https://royalsoftworks.com/download/) [Buy it without a call](https://royalsoftworks.com/pricing/) Half an hour, no slide deck A walkthrough with someone who built the thing. Bring the awkward questions. A pilot in your environment Every feature unlocked, installed with us on the call, configured for your setup. Or just email sales@royalsoftworks.com, answered by a person within one business day. [sales@royalsoftworks.com](mailto:sales@royalsoftworks.com) [office@royalsoftworks.com](mailto:office@royalsoftworks.com) ### Send us a message Tell us what you are trying to do. A person reads it and replies within one business day. Name Email Company— optional Subject— optional Message Send message Goes straight to our own mail server. No CRM, no tracking pixels, no marketing list.