Voiceitt validates next-generation speech AI in new pilot project for the deaf

Summary: Voiceitt has successfully completed a pilot project with deaf and hard of hearing participants, demonstrating the effectiveness and inclusivity of its speech AI technology in recognizing non-standard speech patterns.

Findings:

1. Voiceitt’s technology, which includes a unique database of non-standardized speech samples, demonstrated high accuracy in recognizing and understanding the speech of deaf and hard of hearing people, achieving over 90% accuracy after participants trained the system with 200 recordings.
2. The pilot project underlined Voiceitt’s commitment to inclusivity by adapting the speech recognition software to the specific needs of deaf and hard of hearing users. This improved their ability to interact with voice-controlled devices and reduced their reliance on external interpreters, especially during sensitive or private conversations.
3. The successful pilot project, supported by Newlab, lays the foundation for further strategic collaborations that could amplify the impact of Voiceitt technology and potentially transform the use of speech AI in the deaf and hard of hearing community and beyond.

Voiceitt, an Israel-based provider of speech recognition technology for non-standard language, announced the successful completion of a joint pilot validating its inclusive speech AI technology with deaf and hard of hearing individuals. It demonstrates the flexibility and inclusivity of next-generation speech recognition technology for English-speaking, deaf and hard of hearing individuals with non-standard language.

About Voiceitt technology

Voiceitt’s technology was developed with and for people with non-standard speech, and its unique speech database includes voice recordings from people with impaired and atypical speech to augment and improve Voiceitt’s speech AI models. This database of non-standard speech enables Voiceitt to provide a more comprehensive and accurate experience. Voiceitt further improves its non-standard speech detection by providing personalized detection for people who provide at least 200 recordings.

The pilot was supported by Newlab, a deep-tech venture platform that works closely with innovators as they develop, test and scale cutting-edge technologies. The goal of the pilot was to demonstrate improved accuracy of Voiceitt’s speech recognition on deaf speech and provide qualitative insights to evaluate high-impact use cases for speech AI on deaf and hard of hearing individuals.

“The Voiceitt pilot is in line with the studio’s goals of conducting pilots with startups working on impact-driven advancements in voice services,” says Jamie LePinnet, director at Newlab, who oversaw the pilot. “The success of the pilot lays the foundation for strategic collaborations that can have far-reaching impacts for customers and the community as a whole.”

Although the use of voice-activated smart speakers and other devices has increased, people with non-standard speech due to deafness or hearing loss are often unable to access and use these devices, which pilot participants said “rarely understands their language.” Aside from the convenience of using voice recognition compared to typing, some participants noted that interpreters working with deaf individuals do not always accurately convey what the speaker is trying to express. Another pilot participant explained that people discussing a sensitive topic may prefer not to have an outside interpreter present, and that inclusive voice recognition would help avoid “communication breakdowns” and be better understood.

“Speech recognition technology is incredibly valuable to the deaf community,” says Tyler Pujeda, a Boston-based heart researcher who participated in the pilot and is now an active user of the Voiceitt application. “New inclusive products can make us feel heard and understood.”

Further details on the pilot project

The pilot also analyzed quantitative results from Voiceitt’s proprietary speech recognition software and compared Voiceitt to the industry’s leading speech recognition software providers. Accuracy was measured using word error rate (WER). According to the company, deaf users in this pilot achieved an average of 8% WER (more than 90% accuracy) after training the system on 200 recordings. Voiceitt’s system achieved higher accuracy than leading industry software for the deaf speakers with non-standard speech who participated in the pilot, even without any training.

Dr. Rachel Levy, a speech therapist and customer success manager at Voiceitt, emphasized that while Voiceitt has focused on individuals with a wide range of speech impairments in the past, this pilot project represents a unique opportunity.

“This pilot confirmed Voiceitt’s compatibility and increased accuracy with deaf voices,” says Levy. “But the biggest impact came from the feedback and insights from pilot participants, which led directly to the development of solutions that address the needs and preferences of the deaf community and the other communities we serve and support.”

Photo: Dreamstime