Improvement in mixtures

[nelson c]

Hello Dmitry,
I'm looking at software from my country and found something interesting for the mixes.
There are no fixed times here so the track starts, if not done by the audio level, but in a very smart way.
For the mix point of the current track, the following track is taken into account.
And depending how the sea the beginning of this track, is how it mixed.
The most interesting thing is what it does with the tracks that "explode" in the first seconds.
All the quiet part is played on the track that is ending, and when the track ends, just the part that "explodes" the track that is starting. It sounds very nice and professional. Here is a demonstration.

 

[djsoft]

There's "Trigger mix at" option in RadioBOSS - do you propose also add "Trigger fade in(out) at" as well?

 

[nelson c]

Dmitry,
I know Trigger mix but this is a different approach, the next track will define the mix point of the current one. A picture is worth a thousand words

 

[nelson c]

"Trigger fade in(out) at" as well?

The problem is that tiggermix is different, choose the end point based on the current track (only), this is done by analyzing the next track.

 

[djsoft]

I mean add something like "Trigger Fade In at", "Trigger Fade Out at" and set levels. This appears to be what you describe.

 

[nelson c]

I really do not know if we're talking about the same thing. Fade In and Fade Out are volume fades. That's not what I say at all.

Suppose we have a track "A" that is playing and a "B" that will continue when track "a" ends.

If it detects that track "B" starts quietly and then "an explosive part arrives with a lot of volume" (in the first 5 seconds for example).
The Mix Point of track "A" is modified, making a calculation so that at the moment the track "A" is completely finished, the "explosive part" of track B arrives at that moment. That first part of the Track "B" which is quiet, not deleted, is played simultaneously with track "A"

This same created in sigue editor as the image.
Also attached another demo

 

[djsoft]

Now the picture explains it. So it has to analyze beginning of the following track and adjust the playing track mix point making its loud part start at the end of playing track. This, in some cases, like on your picture, will indeed work better. Noted for the future versions.

 

[nelson c]

Now the picture explains it. So it has to analyze beginning of the following track and adjust the playing track mix point making its loud part start at the end of playing track. This, in some cases, like on your picture, will indeed work better. Noted for the future versions.

Fantastic, I wondered how a local station had such good mixes.
Download the software they use and with several tests I found out he does this.

 

[nelson c]

Any approximation about when the new audio engine will be ready?

I provide more information than I commented on this method of mixing this software that I commented previously:

Its operation is based on what the software calls "point punch in" and "point punch out!
(These points are detected automatically. please see the screenshot one)

Then the player is responsible for reproducing the way you graph in the screenshot number 2.

The result is amazing. In my tests I no longer need to use file types.
The transition between the artist and the music is perfect. even between two songs of different genres.

 

[djsoft]

Any approximation about when the new audio engine will be ready?

This improvement does not require new audio engine. The problem here is how to detect those "punch points" - I suppose in the first incarnation it will be simply level based. Then we'll see how it goes.
This feature will be added after RadioBOSS 5.7 is released, I think it'l l be some sort of mini-beta between 5.7 and 5.8.

 

[nelson c]

I suppose in the first incarnation it will be simply level based. Then we'll see how it goes.

As a first approximation at the entry point I think it does this:
Analyze the waveform and look for a "Abrupt" change in the audio level during its first 10 seconds. That first abrupt change will be the punch in. The previous part will be reproduced in simultaneous on the track that finishes.
In songs that do not have abrupt changes (Because they start very intense, or very quiet) the mixing point is established in the strongest part of the first seconds.

I think that only based on a "fixed" level with some tracks with a low level of audio, important parts of the song can be superimposed.
Anyway it would be a great improvement, we could start with this and see how it goes.

It is great news that will be implemented in the near future.

I will continue to add more information

Thanks

 

[nelson c]

I just realized something else:
When the track starts increasing its level, we need to detect which is the highest part of that increment and establish the punch in there
It is clear that looking at the waveform is where the most appropriate point is, the problem is to create an algorithm capable of detecting it as a human does 

 

[djsoft]

In songs that do not have abrupt changes (Because they start very intense, or very quiet) the mixing point is established in the strongest part of the first seconds.

In this case it can simply mix as it does now - based on time.

It is clear that looking at the waveform is where the most appropriate point is, the problem is to create an algorithm capable of detecting it as a human does 

A simple level-based analysis should do good enough This will be the starting point.

There's already "Trigger mix at" option, but it analyzes the playing track. The new option, instead, will analyze the following track and decide where to put the mix point.