In prisons across the U.S., incarcerated people are being asked to make a strange decision – repeat a series of predetermined words and phrases to an automated voice on a prison phone or lose their phone access in part or in whole. Inmates facing this decision are offered little, if any, information about why they are being recorded or what the recordings will be used for; some inmates are even recorded without being told that it’s happening.
This audio information is being collected and analyzed in order to create voice prints for those inmates being recorded. Voice prints are a unique biometric signature, similar to a fingerprint, that encode characteristics of human speech, like pitch and pronunciation. More than 200,000 inmates have already had voice prints created for them and stored in databases that links their voice characteristics directly to their prison ID number and their other personal information. This type of information is often used for voice recognition purposes, so it’s frequently associated with commercial technologies like Google Home and Siri, but it was originally developed for military and intelligence use. In the context of prisons, voice prints are being used with software (such as Securus’ Investigator Pro) in order to identify the voices of those participating in a call and search for previous calls in which particular voices of interest were present; in short, voice prints are being used for surveillance. In some prison systems, like New York’s, the software is even being used to analyze the voices of people outside of the prison system and track individuals that make regular contact with multiple inmates. This too is often happening without explicit and informed consent from those outsiders being recorded.
Prison authorities and the software companies working with them argue that creating and storing inmate voice prints to be used with voice recognition technologies will help increase prison security and prevent fraud, but rights advocates argue that the collection and compilation of this biometric data has been neither transparent nor consensual.
While there are legitimate security concerns that voice print recognition technology could help address (ex. phone PIN theft leading to fiscal extortion of an inmates family members), there is much preoccupation about the potential future harms that this data collection process could bring both inside and outside the prison context. This scenario is unprecedented, and future legal action have the potential to influence how digital privacy and consent are understood in prisons.
Read more at: https://theintercept.com/2019/01/30/prison-voice-prints-databases-securus/
Data Science Ethics 