Objective: Recently, the Digit Triplet test was shown to be a sensitive speech-in-noise test for early high-frequency hearing loss in noise-exposed workers. This study investigates if a further improvement is achieved when using a closed set of consonant-vowel-consonant (CVC) speech items with the same vowel, and/or a low-pass (LP) filtered version of the standard speech-shaped noise.Design: Speech reception thresholds in noise were gathered for the Digit Triplet, CVC, and CVC_LP test and compared to the high-frequency pure-tone average (PTA). Study sample: 118 noiseexposed workers showing a wide range of high-frequency hearing losses. Results: For the 84 Dutchspeaking participants, the CVC test showed an increased measurement error and a decreased between-subject variation, leading to a weaker correlation with the PTA 2,3,4,6 (R=0.64) and thus a lower sensitivity compared to the Digit Triplet test (R=0.86). However, the use of LP-filtered noise resulted in a sensitivity improvement (R=0.79 versus R=0.64) due to the large increase in betweensubject spread. Similar trends were found for the 34 French-speaking workers. Conclusions: Using CVC words with the same vowel could not increase the sensitivity to detect isolated high-frequency hearing loss. With LP-filtered noise, test sensitivity improved, but it did not surpass the original Digit Triplet test.