Olelo phone captions beat human-assisted ones in FCC testing

In testing for the FCC, MachineGenius's fully automatic phone captions had average error rates of 2.7% to 8.2%, against 8.9% to 19.5% for providers that use human assistants.

  • Not rated a person

    The service had not yet captioned users' calls when the FCC ruled, so the gain is a test result.

Depth runs from 1, a convenience or time saved, to 5, a life or a livelihood preserved.

Status
Intended benefit.
Evidence

foundstudiedreportedclaimed

An independent regulatory document supports the benefit.

Compared with

CA-assisted IP CTS providers, tested on the same call scenarios in 2018 and 2019. Measured.

Without the system
8.9% to 19.5% average error rates
With it
2.7% to 8.2%
The system’s part
MITRE measured the automatic captions against captions from human-assisted providers on the same call scenarios, which isolates the recognizer's output. The assisted providers were tested in 2018 and 2019 and MachineGenius in March 2020. The FCC notes that MachineGenius's agreement with its vendor bars using its calls to "train" the recognizer, which indicates a trained model.
Limits
Not yet checked.
Costs and harms
Not yet checked.
Later
Persisted. Sorenson writes on the app's site that Olelo is now CaptionCall Mobile, part of CaptionCall by Sorenson.

Download data Cite Suggest a correction

Benefit type
Improved Task Accuracy
System
Olelo, MachineGenius's ASR-only Internet Protocol Captioned Telephone Service (IP CTS)
Made by
MachineGenius, Inc. (speech recognition licensed from an unnamed vendor)
Used by
MachineGenius, Inc.
Where
United States
Stage
Tested in a laboratory
Date
, when the source was published
Kind
Aggregate result
Broad benefit
Documenting/Recording Information
Collections
Scrutiny and findings
Last revised
31 October 2026

Sources

  1. Regulatory document, by authors with no tie to the system.

    Federal Communications Commission, Consumer and Governmental Affairs Bureau. Memorandum Opinion and Order, DA 20-485, CG Docket No. 03-123 (conditional certification of MachineGenius, Inc. to provide IP CTS using only automatic speech recognition). Adopted and released 5 May 2020.

    In the recent MITRE testing, MachineGenius’s average word error rates for various call scenarios ranged from 2.7% to 8.2%, while CA-assisted providers’ average word error rates were significantly higher, ranging from 8.9% to 19.5%.

  2. Other, by the makers.

    Sorenson. Olelo Captioned Calls (machinegenius.com). Accessed 31 October 2026.

    An official FCC study rated Olelo — now CaptionCall Mobile — the most accurate captioned call provider.

Details

AI system

Workflow

Automatic speech recognition produced the captions with no communications assistant. MachineGenius chose its recognizer after comparing leading vendors on accuracy, formatting and the handling of numbers. Calls ran over SIP trunks to the Olelo app on users' own phones and tablets.

Follow-up

Conditions for replication

The service must meet the FCC's minimum standards in actual operation. Neither the provider nor its vendor may keep call content after the call.

Entry details

Sector

Telecommunications relay services / accessibility for people with hearing loss.

Purpose

MachineGenius built Olelo to give people with hearing loss captions of their phone calls, produced by automatic speech recognition alone.

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). Olelo phone captions beat human-assisted ones in FCC testing (AIBD-0821). Artificial Intelligence Benefits Database, edition of 31 October 2026. https://petervartanian.xyz/aibd/records/AIBD-0821.html (archived at https://doi.org/10.7910/DVN/AADSII, version 4.0)

BibTeX Data Field definitions