AI-Powered Speech Recognition for Non-Standard Speech #

A company has developed an AI-powered speech recognition system to help people with speech impairments communicate more effectively with both people and digital devices. The technology uses personalized voice models to assist individuals with conditions like cerebral palsy, Parkinson’s, Down Syndrome, or stroke.

The system was initially launched as an app in 2021, functioning as a vocal translator that converted non-standard speech to audio. Users train the AI by recording themselves saying around 200 simple stock phrases.

Originally designed to facilitate in-person communication, the technology has since been adapted for remote work environments. The company has developed integrations with various video conferencing platforms and productivity tools, converting non-standard speech to on-screen captions.

The software is available through per-minute or per-user licenses, with prices ranging from $20 to $50. Bulk licenses can be purchased for workplaces and health or education institutions.

Users have reported employing the technology not only for video meetings but also for writing documents, emails, social media posts, and accessing web browsers by voice. This has expanded digital accessibility for individuals who may have previously faced barriers in certain job roles or in communicating with colleagues or customers.

User Feedback and Future Developments #

A user with muscular dystrophy, who relies on dictation for digital interactions, found the app to be impressively accurate with atypical speech and praised its straightforward training and setup process. However, he noted that professional users might require additional features such as voice-driven cursor control and improved dictation recognition for long-form content.

The user emphasized the need for more comprehensive speech-recognition technology that allows for tasks like managing emails and formatting documents using voice commands. He sees potential for significant advancements in accessibility technology in the future.

Advancements in Speech Recognition #

Recent years have seen notable progress in speech recognition technology. Examples include:

1. A research project collecting voice data from people with various conditions to build algorithms supporting non-standard speech.
2. The launch of an AI-powered feature by a major tech company to recognize a wider range of speech patterns.

Privacy Considerations #

As the technology captures and stores personal data like voice recordings, privacy is a concern for users. The company states that it complies with European Union regulations, which are considered among the highest standards for data privacy globally. User data kept in the company’s database is anonymized, de-identified, and used to improve their algorithm with user consent.

The company aims to not only increase independence in users’ lives and work but also to enable them to enjoy technology and have fun.