Automated Analysis of Relative Fundamental Frequency in Continuous Speech: Development and Comparison of Three Processing Pipelines.
Mark Berardi, Erin Tippit, Yixiang Gao and 2 others
PMID 40348688WHAT IT FOUND
Three automated pipelines for measuring vocal effort in continuous speech matched manual analysis.
The aRFF-B pipeline rejected the fewest samples and is recommended for large-scale use, though vocal fry in female voices remains a major cause of data loss.
Key findings
01All three automated pipelines showed strong reliability compared to manual analysis, with correlation coefficients above 0.80.
02The aRFF-B pipeline had the lowest rejection rates (10% offset, 25% onset) compared to the other automated methods and required the least manual intervention for fricative identification.
03Vocal fry was a primary reason for sample rejection in the aRFF-B pipeline, highlighting a limitation in using continuous speech for female participants.
STILL TO COME
How it was doneWhat they foundWhat it means for SLPs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The study included only female participants, so results may not generalize to male voices. Manual analysis was used as the 'ground truth' for validity, but the authors note this may not be perfectly accurate. A high rate of samples was rejected due to vocal fry, which is common in female speakers and continuous speech, potentially reducing the amount of usable data. The automated pipelines produced fewer usable samples per participant than manual analysis.
Declared interests
The authors have no known conflicts of interest to disclose.
The easy way to misread this
Do not assume these automated pipelines are ready for direct clinical diagnostic use without further validation in diverse populations. The study was limited to female voices and focused on research efficiency; the high rejection rate due to vocal fry means that without specific patient training, a significant amount of recorded data may be unusable.