Remix - Stem Separation

[nivedhya@hotmail.com]

In Remix Window - Stems Separation there is Sensitivity Option given - Default at 0 & varies from -100 to +100.

Where & for What output we have to use this & how?

I tried but unable to notice what is happening?

[stian]

In Remix Window - Stems Separation there is Sensitivity Option given - Default at 0 & varies from -100 to +100.

Where & for What output we have to use this & how?

I tried but unable to notice what is happening?

This changes the sensitivity (or the relative likelihood of detection) for each stem type. So if you for example increase the sensitivity for drums above 0%, there will be more signal components detected as drums.

Best,
Stian

[nivedhya@hotmail.com]

In Remix Window - Stems Separation there is Sensitivity Option given - Default at 0 & varies from -100 to +100.

Where & for What output we have to use this & how?

I tried but unable to notice what is happening?

This changes the sensitivity (or the relative likelihood of detection) for each stem type. So if you for example increase the sensitivity for drums above 0%, there will be more signal components detected as drums.

Best,
Stian

I couldn't get you correctly. I mean, you have said more drum signals will be detected. How is 'more' detection possible than what is present in the mix?

Could you throw more light on this? I mean the same example you said I take, in drum separation if I set 20% plus what will happen,for 50% like so 100% what will be the variation or where should I limit it?

[nivedhya@hotmail.com]

In Remix Window - Stems Separation there is Sensitivity Option given - Default at 0 & varies from -100 to +100.

Where & for What output we have to use this & how?

I tried but unable to notice what is happening?

This changes the sensitivity (or the relative likelihood of detection) for each stem type. So if you for example increase the sensitivity for drums above 0%, there will be more signal components detected as drums.

Best,
Stian

I couldn't get you correctly. I mean, you have said more drum signals will be detected. How is 'more' detection possible than what is present in the mix?

Could you throw more light on this? I mean the same example you said I take, in drum separation if I set 20% plus what will happen,for 50% like so 100% what will be the variation or where should I limit it?

Yes one thing I observed. If I increase sensitivity I could see more of those signals in the output wave form than earlier I mean what were present now little bit blown up so that they are visible.

But What is its use? I mean if I use it in my remix? Will there be any more perceived loudness of that particular stem or exactly What will be the output effect of this improved or increase sensitivity?

[nivedhya@hotmail.com]

In Remix Window - Stems Separation there is Sensitivity Option given - Default at 0 & varies from -100 to +100.

Where & for What output we have to use this & how?

I tried but unable to notice what is happening?

This changes the sensitivity (or the relative likelihood of detection) for each stem type. So if you for example increase the sensitivity for drums above 0%, there will be more signal components detected as drums.

Best,
Stian

I couldn't get you correctly. I mean, you have said more drum signals will be detected. How is 'more' detection possible than what is present in the

Could you throw more light on this? I mean the same example you said I take, in drum separation if I set 20% plus what will happen,for 50% like so 100% what will be the variation or where should I limit it?

Yes one thing I observed. If I increase sensitivity I could see more of those signals in the output wave form than earlier I mean what were present now little bit blown up so that they are visible.

But What is its use? I mean if I use it in my remix? Will there be any more perceived loudness of that particular stem or exactly What will be the output effect of this improved or increase sensitivity?

Neither there is proper elaborate manual nor there is Video Demo like other DAW's & no answer in this Forum if we raise some doubts. After paying & buying should we need such treatment.

Can some relevant people answer for my above Query & we need better support especially for Surround Audio Work

[stian]

In Remix Window - Stems Separation there is Sensitivity Option given - Default at 0 & varies from -100 to +100.

Where & for What output we have to use this & how?

I tried but unable to notice what is happening?

This changes the sensitivity (or the relative likelihood of detection) for each stem type. So if you for example increase the sensitivity for drums above 0%, there will be more signal components detected as drums.

Best,
Stian

I couldn't get you correctly. I mean, you have said more drum signals will be detected. How is 'more' detection possible than what is present in the

Could you throw more light on this? I mean the same example you said I take, in drum separation if I set 20% plus what will happen,for 50% like so 100% what will be the variation or where should I limit it?

Yes one thing I observed. If I increase sensitivity I could see more of those signals in the output wave form than earlier I mean what were present now little bit blown up so that they are visible.

But What is its use? I mean if I use it in my remix? Will there be any more perceived loudness of that particular stem or exactly What will be the output effect of this improved or increase sensitivity?

Neither there is proper elaborate manual nor there is Video Demo like other DAW's & no answer in this Forum if we raise some doubts. After paying & buying should we need such treatment.

Can some relevant people answer for my above Query & we need better support especially for Surround Audio Work

I'm afraid that I'm not quite sure what you are really asking for. I described the sensitivity as good as I can. Please keep in mind that Acoustica merely estimates the stem signals, so you can control the tendency to estimate signal components as either of the five stems. If you don't need it, please just set it to zero.

Best,
Stian

[nivedhya@hotmail.com]

In Remix Window - Stems Separation there is Sensitivity Option given - Default at 0 & varies from -100 to +100.

Where & for What output we have to use this & how?

I tried but unable to notice what is happening?

This changes the sensitivity (or the relative likelihood of detection) for each stem type. So if you for example increase the sensitivity for drums above 0%, there will be more signal components detected as drums.

Best,
Stian

I couldn't get you correctly. I mean, you have said more drum signals will be detected. How is 'more' detection possible than what is present in the

Could you throw more light on this? I mean the same example you said I take, in drum separation if I set 20% plus what will happen,for 50% like so 100% what will be the variation or where should I limit it?

Yes one thing I observed. If I increase sensitivity I could see more of those signals in the output wave form than earlier I mean what were present now little bit blown up so that they are visible.

But What is its use? I mean if I use it in my remix? Will there be any more perceived loudness of that particular stem or exactly What will be the output effect of this improved or increase sensitivity?

Neither there is proper elaborate manual nor there is Video Demo like other DAW's & no answer in this Forum if we raise some doubts. After paying & buying should we need such treatment.

Can some relevant people answer for my above Query & we need better support especially for Surround Audio Work

I'm afraid that I'm not quite sure what you are really asking for. I described the sensitivity as good as I can. Please keep in mind that Acoustica merely estimates the stem signals, so you can control the tendency to estimate signal components as either of the five stems. If you don't need it, please just set it to zero.

Best,
Stian

May be I think I have not put up my question properly. Now let me elaborate. I feed a stereo take out the stems for remix again to my taste. From original input I get some output as Vocals, Drums, etc., Piano etc., This output is original of my input I presume. Afterwards I do some compression or equalization or do something to each of these stems, each stem quality effect I change & remix to my taste.

Suppose while stem separation if I set 30 % to vocals,70 to Piano,0 to Drums what will happen or What is the difference from above stems I get without adjusting in terms of effect?

I mean stem effect changes if I apply some effects like compression etc., in my first case like that altering the sensitivity at stem separation stage itself in what way will affect my Audio?

Having been given What are its advanatages or results in output & in What situations or to my What requirement I have to make use that sensitivity?

I know when to & where What VST effect plug ins to use .Likewise to What output or alter which effect I need to use this sensitivity & in What category it comes.

To my knowledge sensitivity is something related to speakers & Audio Output Quality. Is this too same or something different?

Even you can take this question as a layman asking you regarding this sensitivity you have given in your software & explain to me, Why, What for given, Where to use & after using how to see/observe/feel the resultant output effect.

That is why we ask for Demo Video or Detailed Manual. Have you spoken anywhere in your manual on this sensitivity?

[rafa1981]

Not an expert, but I guess that remix uses machine learning to separate stems, so I guess that you are tweaking the sensitivity/weighting of the algorithm to a given type of sound. Better to mix some drums with a vocal and see how different sensitivities act. The real explanation of what it does might be too technical to be useful.

[nivedhya@hotmail.com]

Not an expert, but I guess that remix uses machine learning to separate stems, so I guess that you are tweaking the sensitivity/weighting of the algorithm to a given type of sound. Better to mix some drums with a vocal and see how different sensitivities act. The real explanation of what it does might be too technical to be useful.

Is there any write up or Article in net (You can give link) to understand what this sensitivity Acon digital talks about so that I can understand?

[stian]

Is there any write up or Article in net (You can give link) to understand what this sensitivity Acon digital talks about so that I can understand?

Are you looking for a mathematical understanding? If that's the case, the sensitivity is a bias before the softmax function described here: https://en.wikipedia.org/wiki/Softmax_function

Best,
Stian

[nivedhya@hotmail.com]

Is there any write up or Article in net (You can give link) to understand what this sensitivity Acon digital talks about so that I can understand?

Are you looking for a mathematical understanding? If that's the case, the sensitivity is a bias before the softmax function described here: https://en.wikipedia.org/wiki/Softmax_function

Best,
Stian

My Question is very simple & still I haven't got a direct answer & only counter questions & round about replies.

1) In stem separation you have given this sensitivity option. To get What result I have to use this option at What setting?
(If I need to use this Where to What requirement how to use and while using so what will be the resultant changes to my stems or those stems in my remix is my simple Query)
2) You said it will detect more signals of a particular stem?
I asked how it will detect more than what is present in the mix but no clear answer?
Or You mean to say normally Acoustica will detect only a portion or certain percentage & if this option is set it will detect more
or
Sensitivity set stem will supress other signals and give out more of a particular set stem.
And now finally you have given a mathematical expression, how do you apply it to Audio signals or here in your Acoustica so it is of no use.
My point straight here is no proper manual for this Acoustica or Video Demo
Simply you put out a demo video & show me using this sensitivity option & its settings & results & give a link. It is more than sufficient than all these round about explanations.
Every software (I don't want to name) have beautiful demo videos or very exhaustive Manual explaining step by step operation which is self explanatory but you take so much pain still we are only debating but .....

[stian]

The point is that the detection algorithm only gives an estimate of the stem signal. It is not 100% accurate, so the sensitivity lets you control the detection if you're not entirely happy with the results. I would recommend 0% as a starting point and leave it there unless there's e.g. vocals leaking through when trying to separate drums. I'm really trying to help you and give as clear answers as I can.

Best,
Stian

[nivedhya@hotmail.com]

The point is that the detection algorithm only gives an estimate of the stem signal. It is not 100% accurate, so the sensitivity lets you control the detection if you're not entirely happy with the results. I would recommend 0% as a starting point and leave it there unless there's e.g. vocals leaking through when trying to separate drums. I'm really trying to help you and give as clear answers as I can.

Best,
Stian

Ok. Thanks for your effort to explain. Sorry it is not to trouble you but to understand it better and make use of it only I raised all those queries.

So can I take from your above statement as conclusion

1) Sensitivity increase the detection of a particular stem by arresting leakage of other stems into that particular stem. If so setting it at max 100 is better if I wish to get that particular stem the maximum?
2) If above is true negative settings less than '0' will means ,it allow other signals to leak more. For example if I set Drums at '-25' will vocals and other signals leak into it? This I'm asking because at some places I need such leakage.
(or)
Is there a set procedure to use this sensitivity setting percentage for getting better/best stems output quality?

Anyhow Thankyou for your patient reply but suddenly halfway, total silence without any replies sometimes back, that too after paying and getting DAW, wishing to use it to maximum but no support made me upset.

And I request please try to support more with demo videos and elaborate manual if possible.

[Tj Shredder]

Its a neural network, a method even the inventors don‘t know exactly whats going on internally. Just experiment and your own neural net, which is in your brain, will get you an idea what it practically means.
Give it some music to separate and listen to the result change the parameters listen again and listen how it differs. No video can give you that experience, its pretty straight forward… The default setting is the best starting point…