Whispp
People who stutter route phone audio through Whispp to convert whispered speech into fluent-sounding voice in real time.
ENABLE Model location
What it is
People who stutter use Whispp to route their phone audio through the application, converting their whispered speech into fluent-sounding voice in real time. Whispp provides an alternate speech pathway that some people who stutter may find more fluent. This tool allows them to communicate more effectively with automated phone systems or other communication interfaces that might otherwise fail to understand their speech. It functions as an assistive technology, which people with disabilities rely on to interact with digital and physical environments when accessibility is not built into a product or service.
Why it matters
The need for tools like Whispp highlights a systemic failure upstream in product design, development, and testing. When mainstream communication systems, particularly speech recognition AIs, are not designed to accommodate diverse speech patterns, the burden of ensuring communication shifts directly onto the user. Assistive technologies, while powerful, often become a necessary patch for inaccessible environments, forcing users to adapt their own methods just to participate. This compensation is required because inclusive requirements were not set, and compatibility with diverse user needs was not adequately tested during pre-launch phases, leading to preventable harm, loss of trust, and diminished autonomy.
Real-world example
A person named Larry, who stutters, was locked out of their online bank account after trying to verify a transaction through an automated fraud prevention phone system. The system's speech recognition AI repeatedly failed to understand Larry's speech, leading to the call being disconnected and the account being locked. Had the bank's system undergone pre-launch accessibility testing with diverse speech patterns, or had an accessible alternative communication channel been provided, this lockout could have been avoided. In such scenarios, individuals like Larry are forced to seek external solutions to bridge the communication gap, enduring significant stress and inconvenience that non-stuttering peers do not experience.
What care sounds like
- “Our speech recognition AI must be usable by people who stutter.”
- “We must perform user studies to test how well our speech recognition AI works for people who stutter.”
- “How can our speech recognition AI do a better job with people who stutter? If it can't, can we provide an accessible alternative?”
- “We designed this system to be inclusive of various speech patterns from the start.”
What neglect sounds like
- “The system works fine for most users, so we assumed it works for everyone.”
- “We'll fix any issues after launch if someone complains.”
- “Only one user reported that issue, so it's not a priority.”
- “Well, it passed our automated tests -- we don't have time to revisit it now.”
- “If users had issues, they should've reported them earlier.”
What compensation sounds like
- “I have to route my calls through this software just to access basic services.”
- “I never know which automated system will be usable today without my special tool.”
- “Why should I need special software to do what others can do out of the box?”
- “I use this application to ensure I can communicate clearly, but it’s an extra step I shouldn't need.”