Pitch tracking improves with new data-driven smoothing method
Pitch Smoothing Using Relative Interval Networks
Sound
Summary
Pitch tracking means identifying the musical note being sung or played at each moment. Existing methods smooth out pitch estimates by looking only at short-term changes and can sometimes get stuck on mistakes that make notes sound an octave too high or low. The authors created Relative Interval Networks, which use a new way to compare pitch differences over longer times and offsets. Their method better corrects these errors and works well across voices and instruments, even when the sound is noisy or unclear.
What this means in practice
- •For speech recognition engineers: Improve pitch tracking in speech systems by reducing octave errors and noise sensitivity during voice analysis.
- •For music transcription developers: Enhance automatic transcription quality by providing smoother and more accurate pitch trajectories for singing and instruments.
Authors
Chin-Yun Yu, Chi-Jen Peng, Li Su, György Fazekas
Abstract
Pitch tracking systems typically couple a per-frame fundamental frequency ($F_0$) estimator with a temporal smoothing stage to obtain continuous trajectories. Conventional Viterbi smoothers enforce first-order continuity but lack long-term temporal awareness and could lock into octave errors across corrupted frames. We propose Relative Interval Networks (RIN), a trajectory smoothing framework that reconciles per-frame pitch estimates with data-driven multi-hop pitch differences. We extract robust relative pitch intervals across arbitrary frame offsets using Variable-Q Transform cross-correlation. We formulate pitch smoothing as an $L_1$-norm optimization problem and prove its equivalence to a minimum cost circulation problem, solved efficiently via linear programming. Evaluations across speech, singing, and instrumental datasets show that RIN substantially improves weak estimators, matches or outperforms Viterbi decoding at a comparable computational cost, and provides superior robustness under certain acoustic degradation.