DTW over WavLM enables text-free L2 speech scoring for phonetics, rhythm, and intonation

The study investigates using Dynamic Time Warping (DTW) on self-supervised WavLM representations to assess phonetic accuracy, rhythm, and intonation in English and Japanese L2 speech without requiring labeled training data. The basic DTW approach comparing learner speech to native templates exceeds human agreement on holistic and sentence-level phonetic scoring.

• For rhythm assessment, the authors introduce methods measuring warping degree in the DTW alignment path, with the best method approaching human-level performance.
• Intonation scoring combines DTW distance over prosodic residuals with pitch and intensity features, though performance remains modest on some tasks.

The results demonstrate that self-supervised representations provide a promising text-free basis for multi-aspect pronunciation assessment.