Could someone with Spectralayer Pro take a look at this audio file for me?

jormungand wrote on 11/27/2016, 5:57 PM This is the link to the audio file i want edited, can anyone with spectralayers pro 3 download a copy of this file, look at it, and edit it so that i can listen to the audio of the two women fighting in the background and mute the volume of the dialogue voices without adjusting the volume of the noises made from the catfight. If anyone is interested then please give it a shot.


irwin-arnstein wrote on 12/15/2016, 4:55 PM

The problem here is going to be that the noise, sound effects and music all are frequencies that cover the "cat" fight dialog in the background. When you have a distinctive noise, in its own notch frequency that is at a high enough volume to hear it clearly, you could start marking it and trying to extract it. At this point, it still requires a human ear to identify dialog that is probably 3 levels deep - i.e. 1) foreground music 2) foreground sound effects 3) foreground dialog and then try to find background dialog. I suppose you could try to remove them in reverse order that I have just talked about with noise inversion layers and then maybe get back to it, but that is a few hours of processing.


jormungand wrote on 12/18/2016, 7:34 PM

Which tools in Spectralayers Pro 3 would you recommend I use to remove the foreground music, foreground sound effects, and foreground dialog?

Layerry wrote on 12/20/2016, 12:57 AM

The harmonics tool and the brush tool are my workhorses. This is going to be a tough one.

DivideFrame wrote on 12/20/2016, 5:37 AM

Your best shot might be to try to extract what you want directly (as Layerry suggested, mainly by using the frequency/harmonics/brush selection tools) instead of trying to remove what you don't want. Keep in mind it's a process based on trial and errors, listen as much as possible the things you select, you can also post-correct your extraction with the eraser tool.

Layerry wrote on 12/21/2016, 2:06 PM

I spent a little time on fhe first part as an exercise for myself. To get this extraction, I created inverse layers for Music, Voiceover, and Attenuation (set volume level to -3db instead of 0db so that it attenuates the sound instead of cancelling it). I set number of harmonics to 40, which takes you up around 8kHz. You'll need to adjust the FFT size for the type of sound you want to work with. Higher FFT (2048) works best for musical tones, lower FFT (1024) works best for noises. Use the harmonics tool for tones. Use the brush tool for noises. I prefer a small brush with a frequency range of 100, time range of 0.10, and hardness of 10% so that I can get very precise with my areas. For large cymbal crashes, I used a larger frequency and time range and painted the area to the Attenuation layer instead of deleting it altogether. The "punch" sound is a combination of descending tone and noise. You can extract it by using the harmonic tool for the beginning tone of the sound, then using the brush only in the low frequencies for the descending tone and the final low thud.