Back to blog

ADAI launches public beta!

·
·
2 min read

ADAI launched its first public beta of a fully-automated audio description (AD) generator. Anyone can now register and upload a test video, and with no need for special knowledge of audio description, download a fully audio described version of the same video.

Offering audio description using AI has been the primary goal of ADAI since its first working prototype in December 2024. Prior to 2024, AD had to be created manually with a team of specialists: people to describe scenes based on knowledge and preferences of visually impaired users who may consumer that particular content, people to edit the descriptions into a script to fit around native spoken audio, people to voice and record the script of descriptions, and people to carefully edit the video so that scenes are described without overlapping the native audio or detracting from the overall experience of the content itself. The human labor involved, while the best available at the time, made creating AD time-intensive and cost-prohibitive. As a results, very little video content was accessible to the visually impaired.

ADAI's beta parallels high-quality work done by traditional, manual AD generation teams, and features a feedback loop to assess quality, a unique feature of ADAI. Feedback is judged based on machine learning as well as user input.

Another feature of ADAI's agentic workflow is customization, allowing people agency over how they want their content described. This feedback came from visually impaired users themselves.

People can customize how much early scene context is generally given, depending whether they choose a profile for creating AD for blind, low vision, or sighted users.

ADAI also offers choice over quality of the voice, whether you want speed and cost savings, or a highly conversational voice with emotive tonality and natural fluency.

ADAI supports choice of over 380 voices across 75+ languages and variants.

And finally, ADAI has a special lexicon dictionary for specifying how particular specialty words or proper nouns should be pronounced, perfect for application in education or specialized industry contexts.

While there are other automated AD generation solutions on the market, ADAI is uniquely built to serve institutions prioritizing quality, security, choice, and personalized customer support.


Share: