Skip to main content

Topic: Tonality estimation? (Read 2256 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.
  • dand
  • [*][*][*]
Tonality estimation?
how to estimate the tonality of frequency components by prediction? if tonal signal is a clean sinusoid (constant amplitude and linear phase) why than iso psycho model II suggests that both amplitude and phase should be estimated as a linear interpolation from two previous blocks? shouldn't amplitude be estimated to be the SAME as amplitude in last block?

  • yanchen
  • [*]
Tonality estimation?
Reply #1
As you said, magnitude estimation from only previous one turly can derive the sinusoid, but it also has the possibility of mistaking another kind of signal which also own the stationary characteristics of magnitude, such as stable background noise.

One or more measurements is the key to credit the judgement. Beside prediction manner, SFM is also popular
for tonal detection.

  • Gabriel
  • [*][*][*][*][*]
  • Developer
Tonality estimation?
Reply #2
I do not know what is SFM, but the good old peak method is quite popular.
FastEnc seems to use a combination of predictivity and peak.