What happened
Hamilton Relay captioned phone calls for people with hearing loss through communications assistants aided by voice recognition. On 9 July 2021, it asked the FCC for approval to caption calls with automatic speech recognition alone. Hamilton told the FCC that its platform draws on several vendors, which can cover for one another. The FCC's National Test Lab, run by MITRE, tested the platform in repeated calls on five scenarios. The platform's average word error rates ranged from 0.9 to 7.7, against 10.7 to 23.3 for human-assisted providers and 4.5 to 17.1 for other automatic platforms. Its captions arrived in 1.0 to 1.6 seconds on average, against 4.0 to 4.8 seconds for assisted providers. On 20 April 2022, the FCC's Consumer and Governmental Affairs Bureau granted conditional certification for two years. On 17 December 2024, it granted five-year full certification for captions by assistants and by speech recognition. It based the grant on Hamilton's annual reports and its actual provision of the service. The FCC names no vendor and does not show that the recognizers learn from data.
Sources
-
Regulatory document, by authors with no tie to the system.
In repeated tests of five call scenarios, Hamilton’s ASR-only platform achieved average Word Error Rates ranging from 0.9 to 7.7, while CA-assisted providers’ average Word Error Rates ranged from 10.7 to 23.3, depending on the scenario.
Hamilton’s ASR-only scores are also generally better than those of other ASR-only platforms, which averaged from 4.5 to 17.1 over the same set of call scenarios.
-
Regulatory document, by authors with no tie to the system.
Based on our review and investigation of the amended application, Hamilton’s annual reports, and other relevant information regarding Hamilton’s actual provision of IP CTS, we find that Hamilton will provide IP CTS in compliance with the applicable mandatory minimum TRS standards.
-
Other, by the makers. Context or contrary evidence.
Hamilton Relay. Hamilton CapTel home page (hamiltoncaptel.com), service notice. Accessed 1 November 2026.
Automatic speech recognition software generates captions of what the other party to the call says, and, in certain circumstances, a live communications assistant may be included on the call to make needed corrections and/or add call details.
Details
AI system
- Workflow
-
Hamilton's platform drew on several speech recognition vendors, so that others could carry the traffic if one failed. Hamilton used the automatic mode as it judged right for users' needs, and users could state a preference. Users could switch from automatic to assisted captions during a call.
Follow-up
- Conditions for replication
-
Neither Hamilton nor its vendors may keep the audio or captions once a call ends. Calls to 911 go mainly to communications assistants, and an assistant joins an automatically answered emergency call as soon as one is free. The service must meet the FCC's minimum standards in actual operation.
Entry details
- Sector
-
Telecommunications relay services / accessibility for people with hearing loss.
- Purpose
-
Hamilton Relay built its ASR-only captioning mode to give people with hearing loss captions of their phone calls without a communications assistant.
30 more questions not yet checked
model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.
Cite this record
Peter H. Vartanian. (2026). Hamilton's automatic captions top assisted ones in FCC lab tests (AIBD-0996). Artificial Intelligence Benefits Database, edition of 1 November 2026. https://petervartanian.xyz/aibd/records/AIBD-0996.html (archived at https://doi.org/10.7910/DVN/AADSII, version 7.0)