updates on the FFT topic.

there is now a node NodeFFT that is capabable of analyzing audioblocks for their frequency spectrum. it is far from being perfect but it does work … basically.

i have added three techniques to improve the quality of the analysis and thereby the detected dominant frequency:

  1. Hamming Windowing is used to shape the signal before transforming it to the frequency domain.
  2. Gaussian Interpolation is used increase the frequency resolution by looking at neighboring bands.
  3. Accumulated Sample Buffer is used to accumulate multiple audio buffers to achieve a higher frequency resolution. since the buffer is filled in a rolling manner the time resolution does not increase in a notable fashion:

example of the first 9 steps populating a 4 times longer Accumulated Sample Buffer with Audio Buffer samples:

1. |(1)|   |   |   |
2. |(1)|(2)|   |   |
3. |(1)|(2)|(3)|   |
4. |(1)|(2)|(3)|(4)|
5. |(5)|(2)|(3)|(4)|
6. |(5)|(6)|(3)|(4)|
7. |(5)|(6)|(7)|(4)|
8. |(5)|(6)|(7)|(8)|
9. |(9)|(6)|(7)|(8)|

note, that the accumulated Audio Buffer sample data would creates audible artifacts, e.g in step 5. where the data of the newly added block connects to a much older block. although this is not correct it does not affect the analysis notably but reduces data handling time significantly.

performance-wise the algorithm is surprisingly fast. using dedicated DSP functions really paid off. the performance can be improved even more by precomputing a Hamming Look Up Table.

Average Duration of Audioblock (μs) on KLST_TINY

TASK μs
no analysis 112.94
no analysis, but update FFT sample buffer 133.90
analysis every beat() 1 133.93
analysis every audioblock() 228.54
  • audio rate ……………………. : 48000Hz
  • samples per audio block ………… : 128
  • max duration of audio block (μs) … : 2667
  • hamming windowing ……………… : true
  • precomputed hamming window ……… : true

  1. note that this measurement does bear much significance as the FFT analysis is performed outside of the audioblock() function but rather in the beat() function. ↩