TY - JOUR
T1 - Giant FFTs for Sample Rate Conversion
AU - Valimaki, Vesa
AU - Bilbao, Stefan
N1 - Funding Information:
The main part of this work was conducted during a research visit of the second author to the Aalto Acoustics Lab between May 30 and June 13, 2022. This research belongs to the activities of the Nordic Sound and Music Computing Network—NordicSMC (NordForsk project no. 86892). For the purpose of open access, the second author has applied a creative commons attribution (CC BY) license to any author-accepted manuscript version arising. The authors would like to thank the Associate Technical Editor for noticing the surprising mismatch between the numerical precisions of the FFT spectrum and the time-domain signal reconstructed using the IFFT. Special thanks go to the anonymous reviewer who suggested testing the method by converting a signal to another sample rate and back.
Publisher Copyright:
© 2023 Authors. All rights reserved.
PY - 2023/3/7
Y1 - 2023/3/7
N2 - The audio industry uses several sample rates interchangeably, and high-quality sample rate conversion is crucial. This paper describes a frequency-domain sample rate conversion method that employs a single large (``giant'') fast Fourier transform (FFT). Large FFTs, corresponding to the duration of a track or full-length album, are now extremely fast, with execution times on the order of a few seconds on standard commercially-available hardware. The method first transforms the signal into the frequency domain, possibly using zero-padding. The key part of the technique modifies the length of the spectral buffer to change the ratio of the audio content to the Nyquist limit. For upsampling, an appropriate number of zeros is inserted between the positive and negative frequencies. In downsampling, the spectrum is truncated. Finally, the inverse FFT synthesizes a time-domain signal at the new sample rate. The proposed method does not produce folded spectral images, like time-domain methods. However, it causes ringing at the Nyquist limit, which can be suppressed by tapering the spectrum and by lowpass filtering. The proposed sample rate conversion method is targeted to off-line use and is useful in audio applications where sound files need to be converted between sample rates at high quality.
AB - The audio industry uses several sample rates interchangeably, and high-quality sample rate conversion is crucial. This paper describes a frequency-domain sample rate conversion method that employs a single large (``giant'') fast Fourier transform (FFT). Large FFTs, corresponding to the duration of a track or full-length album, are now extremely fast, with execution times on the order of a few seconds on standard commercially-available hardware. The method first transforms the signal into the frequency domain, possibly using zero-padding. The key part of the technique modifies the length of the spectral buffer to change the ratio of the audio content to the Nyquist limit. For upsampling, an appropriate number of zeros is inserted between the positive and negative frequencies. In downsampling, the spectrum is truncated. Finally, the inverse FFT synthesizes a time-domain signal at the new sample rate. The proposed method does not produce folded spectral images, like time-domain methods. However, it causes ringing at the Nyquist limit, which can be suppressed by tapering the spectrum and by lowpass filtering. The proposed sample rate conversion method is targeted to off-line use and is useful in audio applications where sound files need to be converted between sample rates at high quality.
UR - https://www.aes.org/journal/
M3 - Article
SN - 1549-4950
VL - 71
SP - 88
EP - 99
JO - Journal of the Audio Engineering Society
JF - Journal of the Audio Engineering Society
IS - 3
ER -