0

I'm working with the aubio library for sound analysis. Basically to detect onset of beats. aubio.onset() takes 4 arguments, 2 of which are buffer size and hop size. I'm trying to figure how changing these 2 values affects my onset detection. 

-B, --bufsize win  The size of the buffer to analyze, that is the length
  of the window used for spectral and temporal computations. Defaults to 512.

-H, --hopsize hop  The number of samples between two consecutive analysis.
  Defaults to 256.

The 2 definitions that I found after doing some online search (https://aubio.org/manual/latest/cli.html#aubioonset) are mentioned above.

From my understanding increasing buffer will require more processing time and increase latency. However is there anything else that buffer size will affect.

Also I did not get what the hop size is, in the above definition what do they mean when they say no of audio samples between 2 consecutive analysis. What is the analysis they are performing and what do they do with the audio samples. How will changing the hop size affect my onset detection

Yash Bhutada
  • 25
  • 4
  • 1
    [This has definitely been asked before](https://stackoverflow.com/questions/51276464/hop-size-meaning-in-aubio/51282039#51282039), at least for hop_size – fdcpp Oct 31 '19 at 20:55
  • 1
    JOS has pretty [extensive online material](https://ccrma.stanford.edu/~jos/sasp/) for Digital Signal Processing as well, if you want a second opinion – fdcpp Oct 31 '19 at 20:59
  • 1
    You’ve pretty much nailed it for buffer size though. Bigger buffer means more detail in frequency domain at the cost of latency. Processing power will be a function of hip size AND buffer size – fdcpp Oct 31 '19 at 21:01

0 Answers0