We are excited to share our unique approach and the preliminary findings on how well virtual assistants can hear us.
RICHMOND, Calif. (PRWEB) April 25, 2018
vocalize.ai is pleased to announce that their vocalyzer test suite has reached the status of MVP complete (Minimum Viable Product). In conjunction with this milestone, the company is releasing a report that outlines their approach and preliminary results. The early results include data collected on the “hearing” assessments for some the most popular voice activated smart speakers on the market. The report also introduces the concept of an “inclusivity index” and outlines how the company plans to apply these tools for detecting algorithmic bias within conversational AI systems.
The vocalyzer test suite has been developed to enable black-box evaluations and benchmarking of AI powered virtual assistants. The goal is to help highlight gaps in performance with measurable, repeatable and shared benchmarks. For the initial task of assessing the hearing capabilities of speech recognizer engines, vocalize.ai has leveraged expertise from the world of audiology; the branch of science that studies hearing, balance and related disorders. Working with talented audiologists, the company has identified procedures and modified them to work with virtual assistants. Audiology tests such as Speech Recognition Threshold (SRT) and Speech In Noise (SIN) are supported by a cloud based utterance database and are readily available for product benchmarking.
“Over the course of last 6 months, vocalize.ai has developed a minimum viable product which is just the beginning of our benchmarking efforts on conversational computing,” states Joe Murphy, CEO & Founder of vocalize.ai. “We are excited to share our unique approach and the preliminary findings on how well virtual assistants can hear us.”
To request a copy of the report, visit the website at http://www.vocalize.ai and select BE VOCAL.