04-26-2021, 10:02 AM | #1 |
Human being with feelings
Join Date: Apr 2021
Posts: 11
|
Remove and replace sound source in ambisonic recordings
Hi,
I would like to remove and replace sound sources in an ambisonic recording. I know blue ripple have a plugin called spatial mask, and also seems to be able to eq a specific area of the ambisonic recording; but haven't either heard or seen its results. Have any of you tried this? Or done it in another way? If so what tools did you use? I am thinking a sort of noise plugin for spatial audio. Afterwards it should be possible to replace it with another sound source with an ambisonic convolution reverb from the acoustic environment. Hope someone have experience with this. Best Jensus |
04-26-2021, 11:47 AM | #2 |
Human being with feelings
Join Date: Jan 2009
Location: Montreal, Canada
Posts: 170
|
You mean replace a sound source in an acoustic recording? That's a task for artificial intelligence, something like what Algoriddim is doing with Neural Mix Pro. The fact that the recording is ambisonic doesn't make the task easier.
|
04-26-2021, 01:44 PM | #3 |
Human being with feelings
Join Date: May 2006
Location: Saskatoon, Canada
Posts: 2,110
|
If you had an anechoic recording and the different sources where far enough apart you might be able to get something approaching what I think you're asking for.
Otherwise the problem is that there will always be some kind of reflected sounds from directions other than the source. These will permeate the recorded channels and make it functionally impossible to remove them. If you want to grab a sound out of an ambisonic recording then you might have success with something like Neural Mix Pro or Zplane PEEL after you have a stereo source decoded from ambisonics to feed to it.
__________________
mymusic http://music.darylpierce.com mywork http://production.darylpierce.com mypodcast https://youtube.com/@ultimatesoundtest |
04-26-2021, 06:33 PM | #4 |
Human being with feelings
Join Date: Apr 2021
Posts: 11
|
Thanks for the quality feedback. The thing is that I am not working with music content, but field recordings. I tried the Peel plugin on a binaural recording, but compared to the demonstration with musical content, where the panning and EQ are set properly, it seems close to impossible to isolate sonic content from a complex sound field. I would like to feed a field recording into the Neural Mix Pro Plugin, but my Mac version doesn't allow me to try the software. Would be fun otherwise, so I must see if I can find a newer model. The idea of combining a spatial and frequency grid in order to remove or reduce a sound source is still a good idea imo, but as you say it is a complex issue and would perhaps rather result in a damping of the original signal and a poor quality of the remaining soundscape.
Software that can analyze which frequencies belong to which sound source seems to be the way for now. |
04-26-2021, 08:07 PM | #5 |
Human being with feelings
Join Date: May 2006
Location: Saskatoon, Canada
Posts: 2,110
|
You are a sound designer ahead of your time. I think such tools will be available at some point but I'm not sure how far down the road. Some of the higher order upsamplers (Sparta, Compass, etc.) might get you to where you can carve a thin enough slice out of the sound field to do something like what you are wanting to. It's kind of like the de-reverb. It works, sort of, depending on the source.
__________________
mymusic http://music.darylpierce.com mywork http://production.darylpierce.com mypodcast https://youtube.com/@ultimatesoundtest |
04-27-2021, 05:31 AM | #6 |
Human being with feelings
Join Date: Apr 2021
Posts: 11
|
Thanks for the links to Sparta, Compass and Aalto University. It seems to be a problem which points toward acoustic scene analysis and stream segregation as well. So I will look into that. Often in noise canceling software you would get a sample of the sound you would like to suppress and make the software lean to recognize its structure. However that isn't quite possible here neither.
|
Thread Tools | |
Display Modes | |
|
|