Sign in

username:

password:



Not a member?

Search Online Books



Search tips

Free Online Books



Chapters

See Also

Embedded SystemsFPGAElectronics
Chapter Contents:

Search Spectral Audio Signal Processing

  

Book Index | Global Index


Would you like to be notified by email when Julius Orion Smith III publishes a new entry into his blog?

  

Minimum Zero-Padding for High-Frequency Peaks


Table: Minimum zero-padding factors $ L_{\hbox {min}}=N_{\hbox {min}}/M$ for keeping peak-frequency bias below approximately $ \Delta $ percent of the sampling-rate divided by the window length [1]. This table is overly conservative for peak-frequencies below 1 kHz. Here, $ N_{\hbox {min}}$ denotes the minimum FFT length, and $ M$ denotes the window length. The zero-padding therefore consists of $ N_{\hbox {min}}-M$ zeros. $ L_{\hbox {min}}$ is calculated using the formulas in [1] and rounded to two significant digits.
Window Type $ \mathbf{\Delta}$ (%) $ \mathbf{L_{\hbox{min}}}$
Rectangular $ 1$ 2.1
Gen. Hamming $ 1$ 1.2
Blackman $ 1$ $ 1.0$
Rectangular $ 0.1$ 4.1
Gen. Hamming $ 0.1$ 2.4
Blackman $ 0.1$ $ 1.8$


Table 4.3 gives zero-padding factors sufficient for keeping the bias below $ 0.01\cdot\Delta\cdot f_s/M$ Hz, where $ f_s$ denotes the sampling rate in Hz, and $ M$ is the window length in samples. For fundamental frequency estimation, $ \Delta $ can be interpreted as the relative frequency error `` $ \Delta f/f$'' when the window length is one period. In this case, $ f_s/M$ is the fundamental frequency in Hz. More generally, $ f_s/M$ is the bandwidth of each side-lobe in the DTFT of a length $ M$ rectangular, generalized Hamming, or Blackman window (any member of the Blackman-Harris window family, as elaborated in Chapter 3).

Note from Table 4.3 that the Blackman window requires no zero-padding at all when only $ 1$% accuracy is required in peak-frequency measurement. It should also be understood that a frequency error of $ 0.1$% is inaudible in most audio applications.5.13


Previous: Optimal Peak-Finding in the Spectrum
Next: Minimum Zero-Padding for Low-Frequency Peaks

Order a Hardcopy of Spectral Audio Signal Processing


About the Author: Julius Orion Smith III
Julius Smith's background is in electrical engineering (BS Rice 1975, PhD Stanford 1983). He is presently Professor of Music and Associate Professor (by courtesy) of Electrical Engineering at Stanford's Center for Computer Research in Music and Acoustics (CCRMA), teaching courses and pursuing research related to signal processing applied to music and audio systems. See http://ccrma.stanford.edu/~jos/ for details.


Comments


No comments yet for this page


Add a Comment
You need to login before you can post a comment (best way to prevent spam). ( Not a member? )