Newbie - Help click-free transition between cues

[mbatchelor]

Hi all,

I'm new to SCS (bought pro), but searching the forum and help hasn't helped me solve my problem. I'm working a musical with 32 numbers. For each number, the director would like cues at rehearsal markings so that she can jump to "measure 39". No matter what I try, there isn't a seamless transition between one cue and another - there's always a click. But there's never a click when I "test" the track. I've turned off the internal mixer, and I've set the number of files to pre-load to 10 (well past my trial sample).

Some things to note before I list my findings - I loaded up the audio file in Audacity so I could find the audio zeros near my measure numbers. I wish I could give more granular numbers - SCS allows you only two decimal places, but in Audacity I can see that the zero-crossing is around the ten-thousandths area (there are zero crossings at 28.42150, and 28.42220, but not 28.42000 for example). Anyway - here's what I've tried:

Attempt 1- Just going cue to cue:
I created the first audio file cue, and copied it 5 times so I could specify the divisions:
0:01.80 - 0:28.42 - Opening - 1 Recitative
0:28.42 - 0:52.74 - Opening - 2 Count-in - M 21
0:51.74 - 1:26.03 - Opening - 3 M 21 - M 39
... etc ...
Result 1- There is an audible click going from "Opening - 1" to "Opening - 2" that doesn't exist when I use "Test" in each cue.

Attempt 2 - Try to fade one out and the other in...
0:01.80 - 0:28.52 - Opening - 1 Recitative (.1 longer than before)
- Set fade-out time to 0:00.10
0:28.32 - 0:52.84 - Opening - 2 Count-in - M 21 (.1 earlier and longer than before)
- Set fade-in time to 0:00.10
- Set fade-out time to 0:00.10
- Set activation to Auto-Start 0.20 seconds before end of Opening 1
... etc ...
This should cause the first one to fade out while the second one is fading in - effectively a cross-fade - right?
Result 2- Even worse - a major stutter now between the cues. And, in case my figuring wasn't correct, I also tried the auto-start thing at 0.10 just in case my double-offset figure wasn't quite correct.

Attempt 3 - I really lengthened it - I decided to overlap them by a whole second - one fading up, the other fading down:
0:01.80 - 0:29.42 - Opening - 1 Recitative (1.0 longer than before)
- Set fade-out time to 0:01.00
0:27.42 - 0:53.74 - Opening - 2 Count-in - M 21 (1.0 earlier and longer than before)
- Set fade-in time to 0:01.00
- Set fade-out time to 0:01.00
- Set activation to Auto-Start 2.0 seconds before end of Opening 1
... etc ...
Result 3: Instead of a cross-fade, the software instead did a whole fade on one and a whole fade on the other. That is, it's like it's ignoring the "start 2.0 seconds before the end" part. Changing the fade type to Linear got me a bit closer, but not by much - it's not behaving like a cross fade.

I then tried to search the forum for "fade" (nope - too small), "cross fade" (didn't like that either) and then "cross-fade" (zero hits). Attempting to RTFM (well, RTFH) to locate cue cross-fades only found looping cross-fades. So I came up empty there too.

I'm rapidly losing the will to live - this particular song doesn't even have any vamps - I can hardly wait until I get to those... I'm certain I'm not the only one to do this. What am I missing?

Thanks for any/all help you can provide,

Marc

[jmgordon]

A quick reply, I haven't tested this. Have a look in "Help" for 'Set Position' Cues in the Editor. As I understand it, the problem is that having chopped up a number to give you alternative start points, you cannot run through the whole track without the clicks. If you leave the track unchopped you could identify the time points (as you have done already) that the director might want to start at, and create 'Set Position' cues to jump there whilst the number is playing (I haven't checked to see if it works if the track is paused in the cue list). It may click when it jumps, but that may not matter.

The cues could be on hotkeys, but if you're going to have up to 5 start points in 32 numbers you'll run out of keys.

I hope that gives you food for thought and experiment.

Malcolm

[mbatchelor]

Thanks for the response. I did try the "Set Position" cues which, pardon me for saying so, is a complete hack. They're presently done by hacking into the description field of note cues. They're insufficient for a few reasons (but thank-you so much for your thoughtful reply - I really appreciate all attempts at assistance). For those that are interested, the following are reasons why the hack is not suitable at this time:

a- Notes can only be a cue, not a sub-cue (although you can completely hose your project by selecting a note cue and clicking the little arrow that will make it a child of another cue - this completely confuses SCS and makes a mess of your notes - so don't try it). Because it's a cue, you can't just skip-over it easily or naturally to go from the Overture to the first song. You have to click past all the note cues to get to the next song.

b- We're operating a non-profit theatre for kids (that's also run by kids). The description in the cue for these things is completely confusing for kids and un-useful - You get the cue name (which cannot have a space in it or the parsing "algorithm" that parses the description gets confused) looks like $SetPos Overture-M21 00:01.23 - not particularly helpful really...

c- You can't simply launch the song at the desired measure. You first have to launch the song from the beginning, then find the correct measure number in the obscure list and right-click on that. So you get the first few measures of the beginning followed by the actual point you needed to start from.

Finally, let me say that if I can split the audio file at 0:28.42 (at a zero-crossing of course) and there's no pop (because it's at the zero crossing), and play it in WinAmp, Windows Media Player, etc without any pops, hesitations, hiccups or glitches, then SCS should be able to do the exact same thing (without the cutting part). I'm defining non-destructive regions to SCS and giving them a name. The bottom line is I have 6 regions (back-to-back) in the same file - playing the cues back-to-back without a pop/click has to be possible. It simply has to be. I just need to know what the magic special-sauce trick is to getting it to work in a deterministic manner (or be told that it's a bug that will be fixed within some kind of timeframe).

I have an advantage here in that I created all the audio tracks that I'm trying to cue up with SCS - so I can see with absolute certainty where there is zero sound (a zero-crossing) and know where it's safe to split, join, or start audio playing.

Thanks all for any incantations that I can wave over the tool that will let me set up these cues with a minimum of fuss.

Marc

[jmgordon]

OK, here's an alternative approach to think about whilst Mike comes back with an explanation about the cue system and drivers used in SCS.

As I understand it, in each number your MD wishes to pre-specify up to 6 or more start points for rehearsal, say "Letter A, B, C etc" or "Bar 29, 45 etc" (I tend to use that rather than "Measure"). In rehearsal she would like to either play the whole number or jump to Bar 29 and play to the end of the number or when she says stop. Your problem is that, if you chop the track into segments 1-28, 29-44, 45-60, 61-end, you have to go over the joins. How about, instead, making your segments 1-end, 29-end, 45-end, 61-end etc. ? You can start where you want and run to the end of the number without clicks.

Set up the show as Q0 Overture, Q1 Song 1(Title of number), Q2 Song 2, Q3 Song 3 etc. till you have all 32 numbers. After Q0 insert Q0.1 Overture bar 29, Q0.2 Overture bar 41, Q0.3 Overture bar 65; after Q1 insert Q1.1 Song 1 bar 44, Q1.2 Song 1 bar 77 etc and so on through all the numbers. It shouldn't take too long if you already have all the edit points. In rehearsal your young people will be able to quickly step through the fully descriptive cue list to find the start point required.
When you get to show time, go through the cue list in the Editor and uncheck "Enabled" for all the decimal cues. They will disappear from the running cue list, which will only include the "whole" numbers. (See description in Help file).

Any good?

Malcolm

P.S. Incidentally, Set Position cues have only just arrived in version 10.6.1. Initially intended for video and image cues, the Help file does make clear that "SCS currently provides an interim or temporary implementation of the 'Set Position' cue type, using the Note cue type. A full implementation of the 'Set Position' cue type will be provided later." Mike works wonders, but miracles take a little longer.

[Mike Daniell]

SCS will not seamlessly play a track that has been split into multiple cues or sub-cues, because it does not pull the audio data into a common buffer. Instead, it will play one file and when that file has completed will play the next file (if that's the way you have set up your cues or sub-cues). So there's no concept of SCS buffering the second file's data hard up against the end of the first file's data.

Conceptually, what you are wanting is a new feature - the ability to set cue points in a file, and a facility to jump to any nominated cue point within that file. Since SCS currently doesn't have that feature, we need to look for the next-best method of achieving this result. Malcolm's suggestions of either using 'set position' cues, or using part-file cues, each of which goes through to the end of the file, are the best solutions at the moment. Note that when you get to show time and disable the part-file cues (as suggested by Malcolm), these part-file cues will not appear at all in the main screen, so the main screen will be 'clean'.

[mbatchelor]

How about, instead, making your segments 1-end, 29-end, 45-end, 61-end etc. ? You can start where you want and run to the end of the number without clicks.

Hey Malcom,

Great idea - thank-you very much. If SCS supported more than one loop per audio-file cue, I think this approach could work. But alas, most of the numbers in the show are filled to the brim with one vamp after another after another.

A typical song looks like:
[Vamp at start] - 5-15 measures - [grand pause or vamp] - rinse and repeat - [vamp/gp] - big closer

It's definitely not something you can account for with a CD player. You *could* do something like it with something like winamp or another player if you create vamp "tracks" that have like 100 iterations - but media players don't understand "vamp" - so you'd have to carefully switch tracks at a measure boundary. SCS has a built-in understanding of looping - so that's perfect.

However, I'm apparently trying to use SCS in a manner it wasn't designed for. I think perhaps that most people are using SCS for augmenting/supporting a show with sound effects, lighting and video, etc. I'm attempting to use the software to actually replace the orchestra altogether. In past productions, I was able to do this pretty simply with a CD player and with the actors understanding that there is no ability to draw a song or fermata out past what they practice with on the CD. This particular show though relies a lot on interminable vamps and huge grand-pauses for large chunks of dialog.

So - if the prevailing answer is that I need to keep doing what I'm doing (working around it with fuzzy .06-.09 delays and overlapping ranges), then I guess I'll keep pressing on. And, I'll submit some enhancement requests to SCS for future consideration.

P.S. Incidentally, Set Position cues have only just arrived in version 10.6.1. Initially intended for video and image cues, the Help file does make clear that "SCS currently provides an interim or temporary implementation of the 'Set Position' cue type, using the Note cue type. A full implementation of the 'Set Position' cue type will be provided later." Mike works wonders, but miracles take a little longer.

I feel like I'm coming in late to a party ... Is this a one-man deal? I thought, with it being version 10.x, that there was support and multi-person engineering infrastructure behind this.

Thanks again for any and all assistance with issues and questions. I'm glad that there's a community of users and experts out here.

Marc

[mbatchelor]

SCS will not seamlessly play a track that has been split into multiple cues or sub-cues, because it does not pull the audio data into a common buffer. Instead, it will play one file and when that file has completed will play the next file (if that's the way you have set up your cues or sub-cues). So there's no concept of SCS buffering the second file's data hard up against the end of the first file's data.

Ok - I think I'm seeing that now.

Conceptually, what you are wanting is a new feature - the ability to set cue points in a file, and a facility to jump to any nominated cue point within that file.

I mentioned in an earlier post that I'm probably trying to use SCS in a manner inconsistent with it's design. This simply confirms it - thank-you.

Note that when you get to show time and disable the part-file cues (as suggested by Malcolm), these part-file cues will not appear at all in the main screen, so the main screen will be 'clean'.

I'll read up on how to disable cues - thanks!

Thanks,

Marc

[Mike Daniell]

If SCS supported more than one loop per audio-file cue, I think this approach could work. But alas, most of the numbers in the show are filled to the brim with one vamp after another after another.

A typical song looks like:
[Vamp at start] - 5-15 measures - [grand pause or vamp] - rinse and repeat - [vamp/gp] - big closer

Sounds like a new feature request! If others are interested in multiple loops per cue, let me know.

I feel like I'm coming in late to a party ... Is this a one-man deal? I thought, with it being version 10.x, that there was support and multi-person engineering infrastructure behind this.

I first developed SCS about 15 years ago after getting frustrated running sound effects using cassette tapes. The theatre directors and many actors and crew started telling me I should sell the program as it was so useful in rehearsals and show production. So after a lot of work on improving the program, I set up a web site and interest in the program gradually grew. Although I'm the only developer, significant design assistance has been provided by many professional and amateur users. For example MIDI Control and Control Send cues were not available in the early versions, but following requests and detailed information received from users these features were added. Two years ago I quit my 'real job' (as a computer systems designer) to work full time on SCS, and more major changes have been implemented since then. I would love to be able to employ another developer, but financially this is currently not feasible.

[dee99]

I've been using SCS since the early days and get a fantastic response and support from Mike and other users. In fact it's better than a lot of large team commercial stuff out there.

I can see where your coming from in regard to regard to multiple queue points in one file and think it would be a good feature to have. There is a problem (correct me if I'm wrong Mike) in that SCS does not decode the audio file but controls the codecs that decode the audio. This means that SCS does not have control over stopping on a zero crossing point without reading the output buffer, plus even if you stop the audio on a zero it can still sometimes sound very abrupt or click like to the human ear depending upon the waveform at the stop point. For MP3 and AAC (sorry I'll have to read up on FLAC) you feed a whole "frame" of data into a codec and all you know about it is that it will start at a certain time and will play for an amount of time.

The solution I have used in the past if to divide the waveform up in Adobe Audition for example and then apply a very fast fade in and fade out to each section (10mS or so), this is fast enough for the human ear not to hear it but has the effect of removing the sharpness/clicks at the start and end of each section. Additionally you could also apply a swept filter over a similar time (10mS) at the start and end of each section starting at low pass to full bandwidth and vice versa.

If you are left with any "holes" between sections that are audible then you can adjust the start time of a cue when auto triggering to be a fraction earlier. For real accuracy on triggering you could use MIDI triggers and a nice fast PC. I have also been known to chop the sections slightly longer than need be (or cut and paste a tiny repeat in) so that I have small sacrificial bits with the same content for the fade in and out parts, bits that small are very hard to hear but are necessary to make it sound correct.

One silly question, what PC are you using? Have you looked at the task manager to make sure its got plenty of performance headroom?
Dee

[mbatchelor]

First Dee, let me thank you for a thoughtful answer to my questions and inquiries! I appreciate any help I can get.

I can see where your coming from in regard to regard to multiple queue points in one file and think it would be a good feature to have. There is a problem (correct me if I'm wrong Mike) in that SCS does not decode the audio file but controls the codecs that decode the audio. This means that SCS does not have control over stopping on a zero crossing point without reading the output buffer, plus even if you stop the audio on a zero it can still sometimes sound very abrupt or click like to the human ear depending upon the waveform at the stop point. For MP3 and AAC (sorry I'll have to read up on FLAC) you feed a whole "frame" of data into a codec and all you know about it is that it will start at a certain time and will play for an amount of time.

By zero-crossing, I'm specifically stating that for both channels (left and right), there is a zero value (i.e. no sound period). In other words, it's the same as a zero-value leader. Finding a zero-crossing (zero amplitude) is a feature of tools like ACID, Cakewalk Sonar, and even the free Audacity - it can help you find an absolute zero crossing where it's completely safe to split a file. It is a guarantee that if you split two audio files at a zero crossing, that you can seamlessly merge them without any click or pop. I think the ultimate issue going on here is one of my requests for enhancement - the ability to be much much more granular than hundredths on start/end/loop times. In Audacity, I can easily find zero crossings at the ten-thousandth or thousandth area. When you really really zoom in on a wave form to see where both channels cross zero, you'll find that it's almost never at a hundredth. Since the underlying BASS libraries in use actually support much more granular start and stop times, I believe I could solve some of these issues by being able to specify 00:00:07.63702 instead of 00:00:07.64 (not even close to the zero-crossing of both channels). Another very useful feature I've requested would be to easily subdivide a single audio cue into multiple slices as cues. If the software "knows" that all sub-cues for an audio-file are using the same file, then all those sub-cues could use the same buffer and there could be a completely seamless transition from one to the other.

Zero Crossing Sources:
http://www.rippletraining.com/zero_cros ... edits.html
http://audacity.sourceforge.net/onlineh ... u_edit.htm
http://www.cakewalk.com/Tips/MixMast/GLooped.aspx
http://books.google.com/books?id=beEFOm ... ls&f=false

The solution I have used in the past if to divide the waveform up in Adobe Audition for example and then apply a very fast fade in and fade out to each section (10mS or so), this is fast enough for the human ear not to hear it but has the effect of removing the sharpness/clicks at the start and end of each section. Additionally you could also apply a swept filter over a similar time (10mS) at the start and end of each section starting at low pass to full bandwidth and vice versa.

I've resorted to the same tactic - basically overlapping windows of audio with fade-in and fade-out. Fortunately I don't have to do a bunch of copy/paste tricks, but there are a couple of vamps that I have where I'm dissatisfied with the clicks I have.

One silly question, what PC are you using? Have you looked at the task manager to make sure its got plenty of performance headroom?
Dee

I've got a Dell Latitude D830 laptop with 4gb of ram. When running cue-to-cue, it's the only thing of significance running. If SCS takes more horsepower than this laptop, I would be surprised. I'm a software engineer (by trade), and I run debugging tools, multiple servers, and multiple clients on this machine whilst performing my day-job activities.

Thanks again for your thoughtful answer. I really appreciate any tips or best-practices hints anyone can provide.

Take care,

Marc