Semi-Automatic Assessment of Vocalization Quality for Children With and Without Angelman Syndrome.
Automated vocal scoring catches tiny speech gains in Angelman syndrome that regular tests miss.
01Research in Context
What this study did
Subramaniam et al. (2023) built a computer tool that scores vocal maturity from day-long home recordings.
They tested it on kids with Angelman syndrome and on typically developing kids.
Two human coders also scored the same clips to check if the computer matched them.
What they found
For typical kids, the computer scores lined up with standard language tests.
For Angelman kids, the scores did not line up.
The tool still agreed well with human coders, so the mismatch is real.
How this fits with other research
Mertz et al. (2014) tracked Angelman kids for 12 years and saw almost no expressive-language growth.
Subramaniam et al. (2023) now show that even when new sounds appear, old tests miss them.
Black et al. (2022) got 92% accuracy classifying autism from voice plus language.
Subramaniam et al. (2023) take the same voice-tech idea but aim it at Angelman syndrome, not autism.
Why it matters
If you serve a child with Angelman, do not trust a low score on a standard language test alone.
Use day-long audio and automated vocal maturity scores to spot small gains the test hides.
Share the audio clips with parents so they can hear progress too.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Record 30 minutes of free play, run the free ULP script, and graph the child’s vocal-maturity score beside last week’s.
02At a glance
03Original abstract
Automated methods for processing of daylong audio recordings are efficient and may be an effective way of assessing developmental stage for typically developing children; however, their utility for children with developmental disabilities may be limited by constraints of algorithms and the scope of variables produced. Here, we present a novel utterance-level processing (ULP) system that 1) extracts utterances from daylong recordings, 2) verifies automated speaker tags using human annotation, and 3) provides vocal maturity metrics unavailable through automated systems. Study 1 examines the reliability and validity of this system in low-risk controls (LRC); Study 2 extends the ULP to children with Angelman syndrome (AS). Results showed that ULP annotations demonstrated high coder agreement across groups. Further, ULP metrics aligned with language assessments for LRC but not AS, perhaps reflecting limitations of language assessments in AS. We argue that ULP increases accuracy, efficiency, and accessibility of detailed vocal analysis for syndromic populations.
American journal on intellectual and developmental disabilities, 2023 · doi:10.1352/1944-7558-128.6.425