The FCC's Consumer and Governmental Affairs Bureau has given conditional certification to the first automatic speech-recognition technology to provide captioned calls for the deaf and hard of hearing.
Getting that go-ahead was MachineGenius. If it pans out, the technology could replace the human assistants now employed for IP Captioned Telephone Service (IP CTS).
In June 2018, the FCC signaled that automatic speech recognition can be a "permissible" way to deliver captioned phone service.
The conditional certification means that MachineGenius can get telecommunications relay service (TRS) funding pending the lifting of conditional status. The FCC has to make sure that the technology does actually meet (or surpass) FCC minimum standards, which it says it expects to be the case following testing.
“So often we have found that technology can provide important solutions to the challenges of ensuring the availability and reliability of accessible communications," said FCC chairman Ajit Pai. "We’re hopeful that this new tool will help bridge the communications divide while meeting our stringent requirements.”
MachineGenius will employ an app that can be installed on WiFi or mobile data-enabled devices.
In what could be called a John Henry vs. the steam drill test, the FCC did comparative performance testing and found MachineGenius' average word error rates were "significantly" lower than those of human communications assistants, as were the average caption delays.
The FCC also said that MachineGenius has established that it is capable of keeping American Sign Language (ASL) conversations confidential and of handling emergency calls within FCC rules.
