Background: Cochlear implant (CI) audio processor upgrades have demonstrated speech recognition benefits in non-tonal language populations, but high-level evidence for native Mandarin-speaking CI users (a tonal language with unique signal processing requirements) remains critically limited. This study aimed to assess the speech perception performance of the SONNET 2 and RONDO 3 dual-microphone audio processors in native Mandarin-speaking CI users.
Methods: This prospective single-subject repeated-measures study enrolled 51 native Mandarin-speaking CI users. Speech recognition performance was tested across five processor configurations: the legacy baseline processor, SONNET 2 (S2) in omnidirectional (S2.OMNI) and adaptive intelligence (S2.Adaptive) modes, and RONDO 3 (R3) in omnidirectional (R3.OMNI) and adaptive intelligence (R3.Adaptive) modes. Outcome measures included monosyllabic words, disyllabic words, and sentence recognition in quiet, and sentence recognition in co-located speech-shaped noise (S0N0 paradigm). The pre-specified primary endpoint was sentence recognition in noise for S2.OMNI vs. the legacy processor; confirmatory linear mixed-effects models (LMMs) and subgroup analyses were exclusively performed for the primary endpoint, with pairwise comparisons for all secondary exploratory endpoints.
Results: In quiet, all four upgraded configurations yielded significantly higher monosyllabic word recognition scores vs. the legacy baseline (all p < 0.05, FDR-adjusted); all configurations except R3. Adaptive showed significant improvements in disyllabic word recognition (all p < 0.05, FDR-adjusted). In the S0N0 noise condition, S2.OMNI and R3.OMNI significantly enhanced sentence recognition vs. the legacy processor (p < 0.001 and p = 0.011, respectively, FDR-adjusted), while no significant benefit was detected for either adaptive mode after FDR correction. LMM analysis confirmed that upgrading to the S2.OMNI configuration was an independent positive predictor of noise sentence recognition (F = 9.885, p = 0.003), with consistent significant benefits across pediatric/adult and unilateral/bilateral users in pre-specified subgroup analyses.
Conclusion: This study provides confirmatory evidence that upgrading to the S2.OMNI configuration significantly improves sentence recognition in noise in native Mandarin-speaking CI users, with consistent benefits across key clinical subgroups. Exploratory analyses show that the S2 and R3 processors also deliver significant improvements in word recognition in quiet, These results fill a critical evidence gap for tonal language CI populations and may help guide clinical device selection.
Keywords: adaptive noise reduction; audio processors; cochlear implants; mandarin Chinese users; speech recognition in noise.