Love it or loathe it, computer-based recording has made compiling the perfect performance a whole lot easier than with that pesky tape. We look at 'comping' techniques in Cubase 4.
Whatever your own personal stance on the relative merits of analogue tape and digital hard drive as alternative formats for multitrack recording, in one regard, at least, computer-based recording wins hands down: the ease of editing. Of course, this can be both a blessing and a curse. On the negative side, it is perfectly possible to overdo the editing process and, eventually, remove the life from a recorded performance. However, used with the right musical sensibilities, the sophisticated editing tools available in a DAW such as Cubase 4 offer a powerful creative environment.
One of the most obvious ways this manifests itself is in the recording and editing of multiple takes. The ability to cycle through a particular song section makes it so much easier for the musician to get into the flow of the performance and, hopefully, produce something memorable. Once the recording has been made, being able to identify the best take, or piece together the 'perfect' take at your leisure, is something that is so much easier in a DAW than with magnetic tape. So, what is the best way to approach this process in Cubase 4?
Cubase 4 's Cycle mode is the most straightforward way to record multiple takes. Setting the positions of the Left and Right Locators to define the section to loop can be done via clicking and dragging the mouse along the main Ruler or via entering the required positions in the Transport Panel (providing you have the Locators section of the Transport Panel displayed). They can also be set to the current project cursor position using key commands ( Ctrl+Number Pad 1 or Ctrl+Number Pad 2 for the Left and Right Locator respectively). Alternatively, the Locators can be set around the currently selected objects (Events, Parts, and so on) within the Project window. Using the 'P' key command simply positions the two locators, while using the Shift+G key combination sets the locators, engages Cycle playback mode and starts playback from the Left Locator — very neat! Of course, it makes sense to set the Locators around a somewhat longer section of the song than you actually need to record in. For example, a couple of extra bars is usually enough to allow musicians to prepare themselves for the start of their performance, while a bar at the end usually allows enough time for any sustained notes to fade out.
When recording audio in Cycle mode, how Cubase actually handles the process can depend upon settings made in two different locations: the Transport Panel and the Preferences dialogue. The choices made in the Preferences / Record window are of the 'set-and-forget' variety. Three options are available for the Audio Cycle Record Mode setting, and the difference between these depends upon an understanding of the differences between audio Regions and audio Events. These are defined in the 'What's In A Name?' box but, for reasons that are explained later, the Create Events + Regions option is probably the most sensible choice to make in terms of flexibility.
If the Record Mode options are displayed in the Transport Panel, a further five settings are available to configure how recording operates in Cycle mode. The first two of these apply to MIDI recording. Of the three options that apply to audio, 'Stacked' is probably the one to choose. 'Keep Last' does pretty much what it says, in that only the last take through the looped section is displayed within the project window (although the other takes are not discarded and can be retrieved from the Pool if required). The difference between 'Stacked' and 'Stacked 2 (No Mute)' is simply that, in the former, all takes expect the last one are automatically muted in the Project window. However, in both cases, the individual takes are all stored on the audio track and, if the Lanes Display Type button is engaged (Lanes Fixed mode), then each take is seen as a separate lane on the track. The last take is placed at the bottom of the stack and therefore forms the currently audible take.
If the Cycle mode setting in the Transport Panel is set to either of the two MIDI options while audio recording is being performed, Cubase defaults to the option specified in the Preferences dialogue (described above) and, again, a separate take is recorded for each pass through the looped section of the song.
With all this preparation done — which takes longer to describe than it does to actually perform — you are now ready to record your multiple takes.
While Cycle mode allows you to endlessly repeat a performance to finally get it right, have you ever been rehearsing a part with the project in playback, only to sing or play the perfect take?
Fortunately, providing that an audio track was armed for recording, that 'I wish I'd just recorded that' feeling can be overcome using Cubase 's 'Audio Pre-Record' option. This allows you to capture up to 60 seconds of audio, even when record was not activated. In order to configure this option, you need to set the length of the pre-record time in the Preferences / Record dialogue. The Cubase documentation says that enabling this option causes audio to be captured into a memory buffer. While no technical details are provided, I assume this does mean a certain amount of RAM is allocated to the task, so there is probably a price to pay in this respect. However, I have not found this setting to have a significant impact upon the performance of my own system and, when you find yourself needing it, this facility can be a complete life-saver.
Having made your recording, it is then time to audition the performances and, if there is not a single stand-out take, to compile the 'perfect' performance by selecting the best bits from each take. There are actually two main (and very similar) ways of doing this: the first, via the Audio Part Editor; and the latter (described below), directly within the Project window. My own preference, for neatness if nothing else, is to use the Audio Part Editor, and I'll look at this first.
In order to use the Audio Part Editor, you first need to combine your various takes into an audio Part. If you have chosen either of the options in the Preferences dialogue that 'Creates Events' in Cycle mode, each take created by the recording processes will be a Cubase audio Event. To combine these into an audio Part simply requires that you select them all (for example, via the mouse) and then choose the Events To Part' option from the Audio menu (or right-click and select the same thing from the Audio / Events To Part option in the context-sensitive menu that appears). The appearance of the takes then alters to indicate the change of status. Double-clicking on the new audio Part will open it in the Audio Part Editor. The appearance of this window is rather like a multi-lane version of the Sample Editor, and a similar set of tools is available.
The first step in compiling the best performance is to divide the individual takes into separate phrases. If the performance has clear periods of silence between musical phrases, this is a relatively straightforward task. If you are happy with a hands-on approach, the Scissors tool can be used to split each take into separate phrases and the periods of silence can be removed. Alternatively — and this is much quicker if you have several long takes to process — you can use the Detect Silence option, by right-clicking on the selected takes and accessing the Advanced options from the pop-up menu. This function includes the ability to define how 'silence' is detected in terms of level thresholds, and also to set a pre- and post-roll time so that phrases do not get cut off too severely. If you select all the takes prior to activating Detect Silence, each take will be processed automatically, making it a very speedy process.
As with lanes on an audio track within the Project window, audio in the very bottom lane of the Audio Part Editor takes priority on playback. However, I find the easiest way to then begin auditioning and selecting from the various takes is to use the Mute tool to mute all the individual phrases, and then set either the Audio Part Editor into Cycle playback mode or the whole project into Cycle playback mode (depending upon whether you prefer to hear the performance you are compiling in isolation or in context). The Mute tool can then be used to switch individual phrases from individual takes on or off as playback is cycled.
Aside from tidying up the occasional location where a phrase from one lane may overlap with that of another, the approach I've just outlined makes it very easy to identify the best bits from each take. In order to reduce the amount of detailed editing required for each phrase, it is worth using the Project / Auto Fade settings to enable automatic fades and crossfades. This applies the settings to all audio tracks automatically and saves you from having to add fades to each and every audio event to get rid of any nasty clicks created by the Detect Silence function. Do bear in mind, however, that these fades are calculated in real time and, if you have a lot of audio tracks in your Project there is a modest CPU impact as a result.
Once you're done, on closing the Audio Part Editor and returning to the Project window you might find that the waveform display in the Audio Part can look a bit strange, as it is based upon the bottom lane of the various takes, and this may include some muted phrases. Visual appearance aside, this is worth putting up with until you are absolutely sure you've finished tweaking the compiled part, and it does make it very easy to move the compiled take within the Project window. However, if preferred, right-clicking on the Audio Part in the Project window and selecting Audio / Bounce Selection creates a new audio Event based upon only the unmuted phrases. This looks much neater but does mean that you cannot go back to editing the compilation in the Audio Part Editor.
The most obvious application for this type of multi-take editing is for vocals but it can, of course, be used with instrumental parts as well. For solo or melodic instrument lines played as distinct phrases, the basic process is identical. However, where one melodic phrase merges into the next, it is obviously more difficult to use Detect Silence to split the takes into shorter sections. A more hands-on approach is therefore required to split the takes at suitable points. Clearly, the best point to place a split is where a well-defined note attack is present. The trickier issue is dealing with the transitions between each phrase, where a fading note in one take may overlap with the attack of a new note in a different phrase. The Crossfades section of the Auto Fades dialogue can help here, and one possibility is to set a standard fade-out curve but design a rapid fade-in, so that the new note takes precedence in the crossfade. If all else fails, some manual editing may be required!
Steinberg use a number of terms to label the audio used in Cubase and it is useful to know what the basic differences are between these. If you'd like a detailed description, Mark Wherry wrote an excellent piece on this very subject in the June 2005 issue of SOS. However, a brief summary of four terms is useful here.
While you are creating audio files when recording in Cubase, the software refers to any audio used within the Project as an audio 'Clip'. Information about Clips is stored separately from the original audio file itself, allowing that file to remain unchanged. However, these audio Clips are presented to the user in three different formats; Events, Parts and Regions. In the context of the ideas presented in the main text, an individual take from a Cycle mode recording session, or a single phrase from such a take once the take has been split, will be an audio Event. If, as described within the main text, several such Events are combined, they form an audio Part (that is, a collection of audio Events grouped together). Regions can be thought of as 'bookmarked' sections of an audio Clip. Essentially, these are sections of audio that you have (knowingly or not!) identified as 'useful' and which you might need to go back to and re-use. Mark explains some of the uses for Regions in the article mentioned above but, in the context of multiple takes, creating Regions from each take allows you to retrieve the original takes from the Pool if you find you need to re-edit them at a later stage of the Project.
If, having compiled the best performance and then moved on to some other aspect of the project, you find you need to return to the original takes for some further work, the Pool can come to the rescue. As you chose the Create Regions + Events option in the Preferences dialogue, each take from the original recording will be readily available within the Pool and can be dragged and dropped back into the project window as required. Once all the takes are placed upon a single audio track, they can be dragged again to recreate the original lane-based display. The compilation process described above can then be repeated.
Of course, it is possible to perform the same splitting and phrase-compilation process with the original audio Events entirely in the Project window. Detect Silence can be used in exactly the same way to split the audio Events into phrases and, with Cycle playback activated, the Mute tool can be used to switch individual phrases from individual takes on and off while the best bits are selected.
Whether you choose to work in the Project window or the Audio Part Editor is very much a matter of personal taste, and exactly the same results can be obtained in either. My own preference is for the Audio Part Editor, as I think this provides a neater solution and makes it easier to flip between editing at the project level and editing at the track level. It also makes some associated tasks (such as moving the compiled take around within the project) a little easier.
Technical issues aside, it is worth thinking about the best artistic strategy before launching into a cut-and-paste strategy to recording a song. This is particularly important with the featured elements of a song, such as the lead vocal, as different singers may be more comfortable (and therefore give a better performance) with different approaches.
While some singers will be more than happy to work on discrete song sections, allowing them to focus in on the delivery of a particular verse or chorus, others may well prefer to run through the whole song, treating it more like a live performance. The latter approach can be particularly beneficial if the singer has performed the song regularly live and is used to delivering the full performance, with associated dynamics.
The other practical consideration is how many takes to record. Again, this is perhaps critical with a singer, and expecting them to simply loop through a song or song section over and over again is eventually going to result in fatigue. If the singer is both well-rehearsed and reasonably competent, then four or five takes is probably going to represent a good compromise. It also gives the producer or engineer (you) a suitable amount of material to go through — fewer takes might not give you the choice that you need, while many more just makes the compilation process very time-consuming. With four or five takes, a quick compilation might be done while the singer takes a short break, before returning to the mic in the same session to re-do any lines or phrases that are problematic. For consistency of sound, this can often be preferable to calling them back at a later date and then having to match the vocal sound.