The spectrograms of the impulse response look identical to me in Praat and Audacity (as far as information is concerned). You sent four pictures, each with a spectrogram in them. On Sunday, Febru2:12:25 PM CET, Boersma Paul via groups.io wrote: Sample I am working on: (dutch commercial utterance for a book) Could anyone give some feedback how to implement this in praat correctly? Specifically is the clap 'too long'? Or is the Formula wrong to compute the inverse in the frequency domain? Other variants such as 1-self produce more recognizable results, but the reverberation is not reduced, rather increased. Another source approaches this visually in the frequency domain by subtraction. I was under the impression that if I would create an 1/value for the real and imaginary part of the audio, this would result in a transfer function that could be applied via convolution. I have an impulse sound (hand clap with a length of 0.17s). Finally, this resulting spectrum is reverse Fourier transformed into sound. Now the spectrum of the signal is divided by the spectrum of the room (i.e. This unwanted filtering can be corrected to some extent by deconvolution. Reversely, the quality of a sound recording can be poor due to reverberation and frequency-dependent absorption by the interior of an inappropriate recording room. I'll try again.įrom "Into Sound" (Ton Wempe, 2018) the following text can be found. The only unanswered post is quite some years old and was not answered. On, at 21:26, stefan via groups.io wrote: I am not sure that this is what you are after, though. If row = 1 then self else -self fi * (self)^2 + self^2) This is because 1 / (a + bi), where a and b are real numbers and i it the imaginary unit, is not 1/a + (1/b)i, but a/(a^2+b^2) - (b/(a^2+b^2))i, so that the formula would be If you are thinking of doing 1 / complexSpectrum, then the formula 1 / self is not the right one.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |