Timefrequency Spectrograms of Orometer Data

1
Time-frequency Spectrogramsof Orometer Data

• Chris Lang
• Indiana University Southeast
• October 11, 2006

2
What this analysis may show

• Frequency of sucking as it changes over time
• Irregularity in sucking, visible as low-frequency
  harmonics
• Jaggedness of sucks, visible as high-frequency
  energy

3
Hai 5-4-06 1400 40-60s

• Strong, regular signal with energy concentrated
  at about 1.3 Hz (a bit more than one suck per
  second)

4
Sma 5-10-06 1500 20-40s

• This shows a change of frequency from above 2 Hz
  down to 1 Hz

5
Mew 5-31-06 1730 (80-100sec)

• This shows a primary frequency of about 1.4 Hz
  and a harmonic frequency of about half of that
  (0.7 Hz), which occurs because the sucks tend to
  alternate in amplitude (tallness)

6
Mew 5-31-06 1730 (140-160sec)

• The sucking in this portion of the signal shows
  less regularity, so the lower frequency
  information (below 1 Hz) is less concentrated.

7
DenB 4-17-06 1630 (80-100sec)

• This shows a noticeable amount of energy in
  higher frequencies (above 2 Hz), which is due to
  the jagged, irregular shape of the suck waveforms.

8
More concerning this analysis.

• Performed using a variant of Fourier analysis
  known as continuous wavelet transform analysis
  which is well-suited for this type of analysis.
• For convenience and for more control, the graphs
  and spectrograms were generated by software
  written by the author but certain commercial
  software programs also will draw similar
  time-frequency spectrograms.
• Acknowledgements Many discussions with Mr. Scott
  Buckley have helped me understand what this
  analysis might say about infant feeding. I am
  also grateful to Scott for providing me with
  Phase I and Phase II data.