Hello,
I am trying to find information regarding SpectralLayers before purchase.
I want to do multichannel recordings, typically 7.1.4 recordings with an ESMA array.
So I will have 11 to 12 channels of audio.
I would like to use SL to isolate specific sounds in all channels at the same time.
Sounds such as birds singing and other specific sounds.
Is this possible in Spectral Layers?
What package would I need?
Hi, and welcome to the forum!
Do you already use another application to create the recording? If that application can host ARA2 plugins, you can integrate SL to the existing process, however multichannel recording is not a feature of SL.
SL can be used for post-processing of existing material and the spectral editing features are IMHO unmatched, however the primary focus of the automated audio separation feature in the current version is music and dialog post-production, so expect to do a lot of manual work to separate nature sounds.
Not in realtime.
I only use SL in ARA2 inside Nuendo, so in Standalone it might be different.
But in my experience, it seems you always process all channels of a multichannel file at once, even if you can set it up to view each one separately. I guess this is to maintain correlation between channels. I mean, if you want to process them separately without regard to correlation, you can easily separate them beforehand.
I think this goes hand in hand with what you want, since you want stuff that is in all channels at once, but that may create problems like a clear birdsong in one channel and a bunch of noise in another. But then you can further process it to clear it of unwanted sounds.
That being said, there is the “Unmix multichannel content” which might be helpful in your case, although I have never used it myself.
There are other unmix tools that might be useful, Like Unmix Levels and Unmix Components, but even without considering the unmix functions, no doubt SL is a VERY powerful ally in doing what you are describing.
I’d get SL Pro trial and give it a go. Read the manual, specially the FFT part if you are not familiar with it, and get used to the various tools and processes, and see where it gets you. Good luck!
1 Like
I record the audio as multicahnnel Wav files on Sound Devices hardware recorder or similar device.
I have seen functions where you can “paint” sounds (such as birds, ambulance sirens and similar) and then isolate the sound together with all related harmonics.
Wouldn’t this work for birds?
I’m not trying to do anything in real-time.
I want to record sounds and then be able to clean them in postprocessing but I want to preserve all of the recorded spatial attributes and spatial cues and maintain the connection between all the channels.
1 Like
Thanks for the input.
I do want to preserve all the time and phase information between the channels, so I want to process them as a multichannel mix.
So if I record a seagul flying over my microphone rig I want to be able to seperate that bird from the background noise.
Does Spectral Layers support full Atmos setups with multichannel recordings?
Can I process 11-12 channels in a spatial audio recording / mix at the same time?
1 Like
Yes you can do that. But initially you’d have information on all channels whenever the seagull is going on, like. The seagull starts faintly on L, for example. Then as soon as the seagull starts on that channel, whatever information is on Ls will start as well, got it? Then for further isolation you can for example go on each individual channel and use any attenuation tool to remove what you don’t want. Attenuation tools are phase coherent, so I guess you won’t have a problem, just an extra step (or two).
Maybe the Unmix Multichannel Content can handle that in an easier way, but I wouldn’t know.
1 Like
You can try the full unlimited capabilities of SpectraLayers Pro free for 30 day by downloading the trial version.
1 Like
Okay! If this is the case then be cautious in mixing in two(or more) audio channels with different sample rates. If you do mix different audio channels with various sample rates and you try to revert(in-terms of flipping the phase) it will introduce irreversible oddities.
As far as the rest of your topic is concerned, search my links that I have posted throughout this forum. I’ve also posted numerous examples on tiktok demonstrating the possibilities of unmixing an old video game(all the sounds and sources, such as gunshots, footsteps, heavy breathing, different voices) and demonstrated that it could be mixed to 5.1 surround sound or even Dolby Atmos 7.1.
1 Like
I will have all channels in the same multichannel wav file, so they will always be the same sample frequency.
One problem I find is that it say it only supports 8ch?
So no height channels in a 7.1.4 setup,is that correct?
Seems strange that Spectral Layers wouldn’t support a full Dolby Atmos setup with height channels?
1 Like
Apparently the amount of maximum channels supported is 16. There are several presets as well as the ability the select a custom channel layout.
Here a screenshot from the New Project dialogue with all possible channels to select from, working in the SpectraLayers Pro standalone application.
To maybe help you with your first question - By default any modification in SpectraLayers is done to all channels (including selections). You do have the ability though to only select specific channels to work on, via the Channels panel.
If you decide to try out SpectraLayers, mind that the channel selection is different from channel Mute / Solo for the purpose of modification (although SpectraLayers will automatically select / deselect channels when changing their audible state, you could still reselect them while maintaining the Mute / Solo). A selected channel is highlighted in the Channels panel.
1 Like
yeah, the vids might look like “painting” sounds
yet, that is likely clone stamp, transform or revealing (mute/ solo/ envelope)
@Unmixing has shown some magic wand screen recording in the past which certainly helped me
still, expecting SLP to unmix birdsong isn’t a module I know of and these are rather manual processes. My workflow for something like that is unmix noisy speech and the unmixed noise will have the birds, other animals, machines, insects, wind
again, modules are typically a starting point, then there is plenty to do manually…and the ability to read the spectrogram is critical when carrying out such manual editing…plus you may well need to develop a workflow to keep track of what you are doing
like others have said, try the demo
I’d like to hear more about this
I’m working in 5.1 right now and, man, it is such a different world to stereo. I’d like to go straight to atmos, but the request I have is for surround…and the cost for atmos monitoring is out of my budget at this time.
So, my main question is how are you recording scenes? On location with cameras? Collecting ambience and reassembling in multi channel? Lots of question down that route.
If you want to remove birds, BOOM library has a specialized tool for that, powered by machine-learning:
1 Like
I isolate birds, insects, etc manually
Magic Wand tool works best for this IME and user needs to get the Magic Wand parameters set just right for manual selections
Yeah, I have done that too. But for the last feature film I worked on the sound, there were dozens of minutes of raw atmo sound. And to get rid of the birds took me a huge amount of time.
1 Like
Interesting, but the demos on their site were not very good in my opinion. Plenty of bird sounds left behind. It seemed to me the plugin had trouble removing the reverberating part of the birds or something. A lot of HF and diffuse bird sounds are still present on the processed version. I suppose it’s less stuff to remove manually after the plugin, tho.
Acon Digital also has a DeBird plugin. Great for removing birds and also for isolating them from background (again, not a foolproof tool).
1 Like
Have you tried using “Select Similar” on each brand of bird? If so, how did it work?
1 Like
I have used select similar on crickets, for which SLP works fairly well, User needs to experiment with settings for best results…can be very time consuming to run the process
cricket sound is very good for testing select similar because the small band freq response and timbre has not much variation
birdsong comes in so many forms; some are a little cloud; some are very unique signature (jay for example)…you’ll have so many different birds it’s a huge task to attempt to find them all
The thing is I don’t want to remove birds, I want to isolate them.
And it’s not only birds, I want to isolate sounds in multichannel recordings in general so that I can build up layers and soundscapes with recordings captured in full 3D, preserve all the spatial attributes but not stack up the noise from each recording when layering them.
I have done some recordings with an ESMA array, following Hyunkook Lee’s recommendations.
I have also used Sennheiser Ambeo mics and other arrays, but I think the ESMA approach is the one that has worked best for me as a universal solution that can also be somewhat portable.
I could not add links to the post but is you search for: “huynkook lee ESMA array” you should get a few useful results.
1 Like