And just because everybody gets this wrong, you can’t do this with one track. A crossfade has two different songs audible at the same time and so the two songs have to be on different tracks. See instructions.
@Steve - why does that page above talk about having to have audio clips (which term it doesn’t explain)? AFAICT it works fine with just selecting across the end/start of audio in a track without any need to add split lines. A primary use case for having this effect in the first place is having a recording of an album on one track (no split lines).
Note to JIMWICK - you can ignore this post because I am talking about features that are not present in the current release version, but will be in the next release.
Are you talking about my recent updates to the alpha manual? If so:
A track may contain several audio clips. The automatic detection of the fade direction is based on the proximity of the ends of the selection to the ends of the selected audio clips, not the ends of the track. So for example, if we have a selection like this:
The automatic setting will apply a fade-in for the first track and a fade-out of the second track like this:
Even though the start of the selection is closer to the start of the second “track”, it is closer to the start of the selected “clip” in the first track. Similarly the end of the selection is closer to the end of the selected clip in the seconds track.
Do we define the term “audio clip” in the manual? If so, we can make the first occurrence a link.
Oh, OK, though if you have complaints about that part of the documentation you should perhaps take it up with the person that wrote it.
The way that I envisage “typical” use of this effect, is that a number of songs / tracks / sound effect … have been imported and the user wants to “merge” them into one track. That is, they want them to run in sequence with crossfades from one to the next.
For example, this image shows two songs:
The tracks can be placed one after the other on the same track like this:
Then you will probably want to cut out some of the “dead space” (the lead out / lead in) between the tracks like this:
Then select the part to crossfade:
Yes you can do that, but I don’t think that is a “typical” use. It has been included for completeness.
If you do have multiple “songs” on a single continuous track, you would probably want to split the recording into separate “songs” (audio clips) so that you can trim out the dead space between the songs.
OK I assumed the author of the effect had written its docs. I certainly need to ask the author if it can be used as many people were asking to use it (on tracks on an album).
I can’t see the point in that unless it’s essential to have only one track or unless you are one of the “challenged” who want to just click and not make any decisions. You may as well use Tracks > Align Tracks > Align End to End then Crossfade Tracks. So I still thought the typical use was on an album.
Note - a crossfade feature with Align End to End would be a time saver.
The reason I was pressing for Crossfade Clips to be included was to help the complainants who could not figure out how to use Crossfade Tracks from the common starting point of multiple songs on one track.
Indeed (unless the inter-track gap is short) you should remove the dead space. I will add that tip. But why would it be mandatory to split the songs to remove the dead space? You can just delete the dead space, no?
Yes, but as Steve says it is “easy”.
If you are doing a crossfade on separate tracks and you slide the second song backwards under the first, so that the one track fades out concurrent with the other fading in, are you not shortening the show?
The only difference is that you did the shortening to your exact specification, rather than the effect doing the shortening to a fixed amount of half the selection.
In my opinion, now that (in Audacity 2.1.3 alpha) we have in “Crossfade tracks” the “Automatic” setting for fade direction, “Crossfade Tracks” is the better effect to use. The main advantage of “Crossfade Tracks” is that it allows the user to precisely set the overlap that becomes the crossfade. However, we have many users that insist on putting their mix into one single track as they work - the Crossfade Clips effect is primarily for those users.
Yes the Crossfade Clips effect can be used if you have a file with several songs one after the other, and using Crossfade Clips is very quick and convenient for this type of task as long as you are not concerned about the “overlap” being precise. In such a case, my workflow would be something like this:
Import the file.
Identify the end of the first ‘song’ / start of the second ‘song’.
Select any silence, or “nearly silent” gap between the songs and “Split Delete”. If there is no noticeable ‘gap’ between the songs, then I’d just add a Split line between the songs.
Select across the split line, roughly 50/50 and apply the Crossfade Clips effect.
Looking at what happens with close attention to technical details:
The manual states:
If the selected region is continuous audio with no splits, the first half and last half of the selected audio will be crossfaded.
If you judge your selection accurately enough, then there is no need for a split line, but if you misjudge, then you probably won’t get what you want.
Let’s say that you misjudge quite badly and select 60/40 (%) rather than 50/50 (this is close to a ‘worst case scenario’ for no splits)
If there is a split line:
The first 60%, from the start of the selection to the split line, will fade out.
The final 40%, from the split line to the end of the selection, will fade in.
The two audio clips are then overlapped such that the fade-out of the first track ends at the same time as the fade-in of the second track ends. (The tracks are always overlapped such that silent end of the longer fade occurs at the same time as the loud end of the shorter fade - this ensures that only the faded regions overlap.)
The second track therefore does not begin to fade in until a short time after the first track has started fading out. However, this ‘error’ is hardly noticeable. The effect is very forgiving in this case.
If there is no split line:
In this case, as the manual says, the first 50% will fade out and the second 50% will fade in.
However, if we have misjudged the mid-point and the first song runs for 60% of the selection, then the first 1/5th of the fade in (which should be the second song, is actually still the first song, and by the time that the second song starts, we are already 20% into the fade-in, so the second song does not fade from silence. This can be demonstrated by crossfading a tone into white noise. In these files (attached), the source material was 5 seconds of 440 Hz tone followed by 5 seconds of white noise. The selection was from 2 seconds to 7 seconds (3 seconds before the end of the tone, to 2 seconds after the end of the tone - a 60/40 selection).
Listen to how the noise fades in.
We should probably have a tip to say that it is generally best to have a split between songs, but I don’t think we should go into detail about all this “under the hood” stuff. Do you think you could work that in somehow?
I haven’t seen those, only the people fading multiple songs in one album track.
Are those who want to put a mix into one track in the Special Interest boards where I don’t look so much?
I would find that hard to describe because it is not easy to me to describe where the fade in should start if the selection is not 50/50.
If more of the tone than noise is in the selection, then after crossfade with split line, I hear the noise starts late compared to if the selection was 50/50. I hear the noise start at about 3s. Why 3s (1/3rd through the result) for a 60/40 selection?
If more of the noise than tone is in the selection, then after crossfade with split line, I hear the tone finishes early compared to if the selection was 50/50.
I also note that when the selection is not 50/50 and there is a split line, then the result is longer than one half of the original selection. But if the selection is 50/50 and there is a split line, the result is half the length of the original selection, just like when there is no split. Right? So where the Manual says
if the selected region is continuous audio with no splits, the first half and last half of the selected audio will be crossfaded.
perhaps it should say something like
if the selected region is continuous audio with no splits, the first half and last half of the selected audio will always be crossfaded, even if the selection includes more of one song than the other.
Then this leads to saying something like using a split line will ensure the crossfade starts or ends at the correct place if the break between the songs is deliberately or accidentally not half way through the selection.
Perhaps you see a better way to put it?
My question would be then, should Crossfade Clips guess by some sort of threshold level where the track break is, so that a split line is not needed?
Yes it should be said that it is safer to have a split line. It would not have occurred to me to add a split line if I deliberately wanted more of one song than another in the selection and then the crossfade sounded wrong.
It’s not usually stated in the topic title, but it quite regularly comes out when trying to clarify what the user’s workflow is. it’s one of the common reasons why some users like to work with lots of projects open - have one track per project and copy / paste from the track one project into the track of another. It’s a ‘valid’ way of working, though imo much less efficient (and probably less safe) than working with multiple tracks in one project. I do suggest that they try using Audacity’s muli-track capability, but some seem to have difficulty getting their head round that way of working.
What I find difficult is to describe it without looking at it in terms of separate tracks, but for those that find it difficult to visualize working with multiple tracks, I don’t think that kind of explanation will help much.
I’ll have a go at making some illustrations.
Nicely put - That’s the crucial point that we need to lead up to.
Perhaps something along the lines of:
If the selected region is continuous audio with no splits, the first half and last half of the selected audio will be crossfaded. When crossfading song this way, the effect has no way of knowing where one song ends and the next song starts, so it assumes a half-way point and makes the length of each fade exactly 50% of the total selection length. If you accidentally or deliberately include more of one song than the other in the selection, then one of the fades will include a portion of both songs, and the crossfade will not be entirely smooth. Using a split line will ensure the crossfade starts or ends at the correct place.
I don’t think there is any reliable way of doing that. Looking for the lowest amplitude would be the obvious approach, but we know how unreliable “Silence Finder” can be, and even my best attempts at improving that are fallible. There’s also use cases such as crossfading across a glitch, where the “mid point” should be the ‘loudest’ part of the glitch.
As I’ve used the effect, I’ve also discovered some new “tips and tricks” that I’d not thought of when I wrote the effect. When I have time I’ll try to work them into the manual.
Do these make the difference any clearer?
They are intended to show how the tracks will fade (“constant power” fade) for two sequential songs - the first song is shown as blue shading, the second song as green shading. The red dotted line outlines the fade before they are overlapped by the effect into a crossfade.
I am not fully understanding “one of the fades will include a portion of both songs”. Even with deliberately selecting more of one song than the other, isn’t the point of the crossfade that at least somewhere, both songs will sound together?
Perhaps to be more clear you could post your blue and green images of split and no-split after the crossfade, but I am still not sure if those images should be in the Manual.
If correct, I think it’s an important point that a split-line crossfade with a non-equidistant selection will produce a selection longer than half the original length.
Overlap the two fades and mix them (‘sum’ them) together.
Steps 1 and 2 may be in any order - you can do step 2 then step 1 or step 1 then step 2 - but the fade-out should ideally contain only the end of the first song and the fade-in should contain only the start of the second song. That is what happens if there is a split line, as can be seen in this illustration:
It should be only when the two fades are mixed (step 3) that both songs sound together. After step 3 you should hear the second song fade in from silence while the first song is fading out to silence.
I don’t hear that in your demo case in the complete cross-faded selection, where selection before crossfade with split line is from 2 seconds to 7 seconds (3 seconds before the end of the tone to 2 seconds after the end of the tone).
After crossfade with split line, I don’t hear the noise at all until about 3s. Spectrogram seems to confirm that.
So to my hearing, if the selection is not equidistant between songs, there will be a substantial area in the selection after crossfade where only one song is playing. Don’t you agree?
So although I can see what you mean by “one of the fades will include a portion of both songs” (if you are referring to before the crossfade) I think it confusing. I would find it clearer to just say “Using a split line will ensure the crossfade starts or ends at the correct place.”