Skip to main content
Signal Processing

Return to Answer

added 92 characters in body
Source Link
Bob
Bob
  • 666
  • 3
  • 6

Do you know the exact duration in samples of the sleep time? If so, then you can take the un-windowed fft of the 2nd block of N samples, modify the phase of each bin to “undo” the effect of the sleep time; then you can inverse transform this back into the time domain and concatenate the resulting data with the data from the first frame. Then you can take ina single length-2N fft (windowed or not, your choice). This assumes the spectrum of the underlying signal is stationary over the time interval 2N+M. The phase modifications should insure no discontinuity when the 2 frames are concatenated.

Do you know the exact duration in samples of the sleep time? If so, then you can take the un-windowed fft of the 2nd block of N samples, modify the phase of each bin to “undo” the effect of the sleep time; then you can inverse transform this back into the time domain and concatenate the resulting data with the data from the first frame. Then you can take in single length-2N fft (windowed or not, your choice). This assumes the spectrum of the underlying signal is stationary over the time interval 2N+M

Do you know the exact duration in samples of the sleep time? If so, then you can take the un-windowed fft of the 2nd block of N samples, modify the phase of each bin to “undo” the effect of the sleep time; then you can inverse transform this back into the time domain and concatenate the resulting data with the data from the first frame. Then you can take a single length-2N fft (windowed or not, your choice). This assumes the spectrum of the underlying signal is stationary over the time interval 2N+M. The phase modifications should insure no discontinuity when the 2 frames are concatenated.

deleted 242 characters in body
Source Link
Bob
Bob
  • 666
  • 3
  • 6

Just a hunch here; you might want to search the term “enhanced spectrogram”. In this techniqueDo you get an estimate ofknow the exact frequency of each fft peak by looking at the rate of changeduration in the phase from 2 slightly offset fft frames (or alternatively by doing a cubic interpolation usingsamples of the upper and low fft neighbors). Ifsleep time? If so, then you were to do this forcan take the firstun-windowed fft of the 2nd block of N samples, you should be able to computemodify the exact numberphase of cycles (integer + fractional)each bin to “undo” the effect of that sinusoidal component thatthe sleep time; then you will be “missing” during your “sleep time”can inverse transform this back into the time domain and concatenate the resulting data with the data from the first frame. Then when Then you can take the 2ndin single length-2N fft, you should know the phase jump that was incurred during the sleep interval (windowed or not, and compensate for it? Thisyour choice). This assumes the underlying spectrum of the underlying signal is stationary betweenover the 2 fft’s.time interval 2N+M

Just a hunch here; you might want to search the term “enhanced spectrogram”. In this technique you get an estimate of the exact frequency of each fft peak by looking at the rate of change in the phase from 2 slightly offset fft frames (or alternatively by doing a cubic interpolation using the upper and low fft neighbors). If you were to do this for the first fft of N samples, you should be able to compute the exact number of cycles (integer + fractional) of that sinusoidal component that you will be “missing” during your “sleep time”. Then when you take the 2nd fft, you should know the phase jump that was incurred during the sleep interval, and compensate for it? This assumes the underlying spectrum is stationary between the 2 fft’s.

Do you know the exact duration in samples of the sleep time? If so, then you can take the un-windowed fft of the 2nd block of N samples, modify the phase of each bin to “undo” the effect of the sleep time; then you can inverse transform this back into the time domain and concatenate the resulting data with the data from the first frame. Then you can take in single length-2N fft (windowed or not, your choice). This assumes the spectrum of the underlying signal is stationary over the time interval 2N+M

Source Link
Bob
Bob
  • 666
  • 3
  • 6

Just a hunch here; you might want to search the term “enhanced spectrogram”. In this technique you get an estimate of the exact frequency of each fft peak by looking at the rate of change in the phase from 2 slightly offset fft frames (or alternatively by doing a cubic interpolation using the upper and low fft neighbors). If you were to do this for the first fft of N samples, you should be able to compute the exact number of cycles (integer + fractional) of that sinusoidal component that you will be “missing” during your “sleep time”. Then when you take the 2nd fft, you should know the phase jump that was incurred during the sleep interval, and compensate for it? This assumes the underlying spectrum is stationary between the 2 fft’s.