Caption Companion Demonstrated for HLAA: Real-World Offline Captioning, Explained
Premier Visual Voice recently presented Caption Companion to members of the Hearing Loss Association of America (HLAA) during a live virtual meeting hosted by the HLAA North Bay Chapter. The session included a full product walkthrough, background on how Caption Companion was created, and an extensive Q&A covering real-world use cases—from healthcare appointments to conferences, classrooms, restaurants, and disaster response.
This blog summarizes the entire discussion, including the most important questions and answers, so readers can understand exactly how Caption Companion works, what problems it solves, and where it fits alongside human captioning services.
Caption Companion was developed by Premier Visual Voice, a company founded in 2012 by professional real-time captioners. Unlike many captioning products created by general technology companies, Caption Companion was built by people who have spent decades providing CART (Communication Access Realtime Translation) and accessibility services.
Margaret Downs, Executive Director of Premier Visual Voice, is a registered CART provider who continues to caption professionally today. Her experience delivering captions in classrooms, workplaces, conferences, and public venues revealed a recurring problem: caption accuracy often fails because of poor audio—not because the captioning technology is bad.
Mathew Call, Director of Technology, is a certified interpreter and captioner with deep ties to the Deaf community. His background includes sign language interpreting, voice writing (re-speaking), and managing remote captioning services long before video conferencing became common.
Together, their shared experience led to a clear realization:
Accessibility technology should remove friction—not create it.
Caption Companion was designed to eliminate the technical barriers that regularly undermine captioning quality.
Caption Companion is not an app. It is a dedicated captioning device—an 8-inch tablet purpose-built exclusively for real-time captioning.
Once powered on, the device begins captioning with a single tap.
Using phones, laptops, or general-purpose tablets introduces common problems:
Caption Companion avoids these issues by being single-purpose hardware with audio quality at the center of its design.
A major theme of the HLAA presentation was that caption accuracy depends more on audio quality than on the speech-to-text engine itself.
Caption Companion prioritizes audio capture through:
By delivering clean audio directly into the captioning engine, Caption Companion achieves faster and more accurate captions than many cloud-based tools—even in challenging environments.
One of the most significant milestones discussed during the HLAA session was Caption Companion becoming the first AI-based captioning solution to pass the NCSP certification exam.
The Global Alliance for Speech-to-Text Captioning administers the National Court Reporters Association-style certification exam used to evaluate professional human captioners.
Unlike typical AI accuracy claims based on word-matching percentages, the NCSP exam evaluates:
This matters because captions that are technically “accurate” at the word level can still change meaning if punctuation or context is wrong (for example, “can” vs. “can’t”).
Caption Companion was tested using the same materials and standards applied to human captioners—and passed.
During the presentation and Q&A, the team emphasized that Caption Companion is designed for real-world, everyday use.
Because it works offline, Caption Companion is especially valuable in environments where internet access is unreliable or unavailable.
Healthcare use was one of the most discussed topics during the Q&A.
Caption Companion is not marketed as HIPAA-certified, because it functions like other assistive communication technologies (such as assistive listening systems or speech-to-text displays), not a medical record system.
However, the device is highly privacy-protective by design:
For personal use, this often makes Caption Companion more private than cloud-based captioning apps that send audio and text to external servers.
Yes. Patients can bring Caption Companion to medical appointments and use it as a personal accessibility tool—similar to how hearing aids or assistive listening devices are used.
Users may also copy text manually to another device (within the auto-delete window) if they want to retain instructions or notes for personal reference.
HLAA members asked extensively about restaurants, health fairs, and other noisy settings.
Users reported successful captioning in:
As with any captioning solution, mic proximity matters, but Caption Companion performs notably well when audio is captured correctly.
Another key Q&A topic was captioning speakers with heavy accents.
Caption Companion supports:
Users can switch dialects in real time to improve accuracy—for example, selecting an Indian English model when listening to speakers from India.
The AI engine performs particularly well with accented speech compared to many live auto-captioning platforms, especially when paired with high-quality audio input.
Premier Visual Voice emphasized that Caption Companion is not meant to replace human captioners in every scenario.
Caption Companion includes the ability to connect to Wi-Fi and display human CART captions when needed, allowing users to choose the right solution for each situation.
Each Caption Companion package includes:
Optional accessories include:
Battery life averages ~11 hours of active captioning, with longer standby time and standard USB-C charging support.
Throughout the session, attendees repeatedly noted:
For many, Caption Companion represented a shift from reactive accommodations to proactive access.
HLAA CA Youtube
To explore specifications, demo videos, and purchasing options, visit the official Caption Companion product page from Premier Visual Voice.
Caption Companion was built by captioners, tested against human certification standards, and designed for real-world accessibility—wherever communication happens.
