Sampling-Frequency-Independent Audio Processing

Proposed SFI convolution layer converting latent analog filters into sampling-frequency-dependent discrete weights — Sampling-frequency-independent convolution layer (Fig. 1(c)). Saito et al., EUSIPCO 2021

We develop sampling-frequency-independent convolutional layers and neural analog filters, including stride conversion, to enable audio models to operate at sampling frequencies not seen during training. Source separation is a primary application. Related work investigates multi-sampling-frequency naturalness MOS prediction using an SFI layer and sampling-frequency robustness in acoustic scene classification through kernel-perturbation resampling.

Demo

Related Publications

Kanami Imamura, Tomohiko Nakamura, Kohei Yatabe, and Hiroshi Saruwatari, “Dissecting performance degradation in audio source separation under sampling frequency mismatch,” in Proceedings of IEEE International Conference on Acoustics, Speech, and Signal Processing, May 2026, pp. 15832–15836.
Kanami Imamura, Tomohiko Nakamura, Kohei Yatabe, and Hiroshi Saruwatari, “Continuous function approximation of convolutional kernels for sampling frequency adaptation of pre-trained source separation networks,” in Joint Meeting of the Acoustical Society of America and the Acoustical Society of Japan, Dec. 2025.
Go Nishikawa, Wataru Nakata, Yuki Saito, Kanami Imamura, Hiroshi Saruwatari, and Tomohiko Nakamura, “Multi-sampling-frequency naturalness MOS prediction using self-supervised learning model with sampling-frequency-independent layer,” in Proceedings of IEEE Automatic Speech Recognition and Understanding Workshop, Dec. 2025.
Kanami Imamura, Tomohiko Nakamura, Norihiro Takamune, Kohei Yatabe, and Hiroshi Saruwatari, “Stride conversion algorithms for convolutional layers and its application to sampling-frequency-independent deep neural networks,” Signal Processing, vol. 242, Nov. 2025.
Kanami Imamura, Tomohiko Nakamura, Norihiro Takamune, Kohei Yatabe, and Hiroshi Saruwatari, “Local equivariance error-based metrics for evaluating sampling-frequency-independent property of neural network,” in Proceedings of European Signal Processing Conference, Sept. 2025, pp. 276–280.
Kanami Imamura, Tomohiko Nakamura, Kohei Yatabe, and Hiroshi Saruwatari, “Neural analog filter for sampling-frequency-independent convolutional layer,” APSIPA Transactions on Signal and Information Processing, vol. 13, no. 1, e28, Nov. 2024.
Kanami Imamura, Tomohiko Nakamura, Norihiro Takamune, Kohei Yatabe, and Hiroshi Saruwatari, “Algorithms of sampling-frequency-independent layers for non-integer strides,” in Proceedings of European Signal Processing Conference, Sept. 2023, pp. 326–330.
Koichi Saito, Tomohiko Nakamura, Kohei Yatabe, and Hiroshi Saruwatari, “Sampling-frequency-independent convolutional layer and its application to audio source separation,” IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 30, pp. 2928–2943, Sept. 2022.
Koichi Saito, Tomohiko Nakamura, Kohei Yatabe, Yuma Koizumi, and Hiroshi Saruwatari, “Sampling-frequency-independent audio source separation using convolution layer based on impulse invariant method,” in Proceedings of European Signal Processing Conference, Aug. 2021, pp. 321–325.