Improving respiratory disease detection through SSL-enhanced acoustic analysis and exercise-rest measurements.

For our peak-performing configuration, the acoustic + wav2vec 2.0 + WavLM fusion model post-exercise, we obtained highly significant empirical p -values of p = 0.00280 for the vowel phonation task and p = 0.00300 for the voluntary cough task, indicating that these results are statistically robust.

also in 132,142 other papers