Audio processing

Hi. Is anybody interested in spectrum processing?

Yes.

Hi. I wish to talk how you process spectrum, I use it to make some retouch to old files: https://youtu.be/wQytDSOqEuc

Do you have a question, or do you want to tell us something, or is there something specific that you want to discuss?

The ringing artefacts are quite bad.

Hello, Steve. I know that curent state of it it is not perfect, I shall try to use neural networks. If you are interested in this you could send me private message. Thank you

Better to discuss on the open forum so that anyone else that’s interested can also contribute.

I plan to use neural networks, FANN

They have two forums here: Fast Artificial Neural Network Library / Discussion

I struggle now with libfanndouble, probably linux library doesn’t work on windows, because earlier i builded fann with this library

*With the other library: fanndouble.dll

If you wish to discuss the fann library specifically, please use the fann forum: Fast Artificial Neural Network Library / Discussion

Okay, thank you. I have figured out what to do but now the results are far from perfection, though they are quite promising

What are the advantages of Nyquist processing?

Compared with what?

Some big advantages for Audacity user of Nyquist compared to MatLab, Python, SoX, any other external programming language:

Nyquist also provides a large number of DSP primitives that makes it extremely simple to write many types of filters, processors and generators, with very little code. As a quick example, to generate band limited noise with a 6 dB/octave roll-off with -3dB frequency corners at 100Hz and 1000 Hz:

  • Using C/C++ it would probably require more than 50 lines of code,
  • In Nyquist it can be written as:
(lp (hp (noise) 100) 1000)

Steve wrote:

Using C/C++ it would probably require more than 50 lines of code,

And probably another 200-300 for headers, entry points etc once you want to turn it into a vst for example.