Neo Scribe
Real-time AI captioning that understands Canadian English, French, and Québécois.
Neo Scribe is our AI speech recognition model, tuned for the dialects and registers Canadian events actually require — including Québécois, where off-the-shelf engines fall short. 98% accuracy, sub-300ms latency, and nothing retained after your session ends.
How It Works
Capture
Audio enters from your venue's existing infrastructure — microphones, mixing boards, streaming feeds, or platform integrations.
Recognize
The model transcribes speech in real time, handling shifts between languages within the same conversation across 36+ supported variants — and identifying multiple speakers without manual setup.
Refine
Domain-specific vocabulary (legal, medical, parliamentary, financial, technical) is loaded before the event so terminology, names, and acronyms resolve correctly.
Deliver
Captions stream to connected displays, AV systems, or directly into Neo Translate, Neo Access, and Neo Connect.
Technical Specifications
98% speech-to-text accuracy measured across real-world deployments
Automatic speaker identification — distinguishes multiple speakers without manual setup
36+ languages including English, Canadian French.
Self-serve vocabulary customization for domain-specific terminology: legal, medical, parliamentary, financial, technical
Automatic punctuation, formatting, and paragraph segmentation
Sub-300ms latency from speech to caption display
Real-time bidirectional audio streaming for continuous, uninterrupted captioning
Noise-resistant processing optimized for live event environments with ambient sound
Zero persistent data retention — audio processed in real-time, deleted when the session ends
Never used for AI model training — your data remains your data
Use Cases
Parliamentary Committee Session
In a committee hearing, an inaccurate transcript becomes a misquoted minister, a flawed Hansard entry, or a corrected public record. Neo Scribe handles formal register, multi-speaker dynamics, and pre-loaded legislative vocabulary — including in Québécois — so the record matches what was actually said.
Medical Conference
Mispronounced drug names and misheard clinical terms aren't typos; they're liability. Neo Scribe's domain vocabulary loads pharmaceutical names, clinical terminology, and diagnostic language before the session — sustaining 98% accuracy through dense, fast medical speech.
Corporate AGM
500+ attendees, multilingual shareholders, and a meeting record that may become a regulatory document. Neo Scribe streams captions to venue screens via Neo Access and drives multi-channel distribution via Neo Connect — without retaining any of the conversation after the meeting closes.
Integration Points
Feeds directly into Neo Translate for real-time caption translation
Displayed through Neo Access on any AV infrastructure in the venue
Managed by Neo Connect for large-scale multi-channel deployments
Frequently Asked Questions
Yes. Automatic speaker identification distinguishes speakers without manual configuration — including in fast-paced, multi-party discussions like committee hearings or panel debates.
36+ languages, including Canadian English, Canadian French. Most international STT engines are trained on generic North American English — Neo Scribe is built for the dialects Canadian events actually require.
Neo Scribe accepts audio from microphones, mixing boards, streaming feeds, and major conferencing platforms. Captions output to displays, AV systems, or downstream Neo products. No proprietary hardware required.
Noise-resistant processing is built into the model. Neo Scribe is optimized for live event environments — ambient sound, room reverb, overlapping speech — not the studio conditions most STT engines assume.
Processed in real-time and deleted when the session ends. Nothing is retained. Nothing is used for AI training.
See Neo Scribe in Action
Schedule a demonstration to see real-time captioning at 98% accuracy with your content.