Scoring Central is a forum hosted by Mattias Westlund, creator of Sonatina Symphonic Orchestra. The forum is currently a small, friendly and supportive community. I can’t come up with a better description than that provided by Mattias himself so I’ll just quote him:
Scoring Central is a forum for all virtual orchestration enthusiasts; hobbyists, amateurs, semi-professionals and professionals alike. Here you can discuss everything from composing and music theory to sample libraries, plugins, DAW’s, audio interfaces, sampling, mixing and lots more. You are also welcome to post your own tracks in the User Music subforum.
I’m a moderator there and I visit the forum daily. If you want to discuss virtual orchestration, sampling, plugins, anything to do with making orchestral music using your computer along with free or low cost sample libraries and free or low cost VST plugins, the Scoring Central forum is a good place to do that.
Proper gain staging, as shown in the video below, has probably made the biggest improvement to the quality of my mixing since I first started. My early mixes (none of which exist anymore) tended to be muddy and I always struggled to make each instrument heard because I kept running out of volume. Once I started following the steps (and the video) below, I started treating mixing as a completely separate step to be started after all composing was complete. Based off what I learned from the video, and my own trial and error, my process is as follows:
While composing, I go ahead and adjust volume faders as needed and get a rough mix just so that I can get an idea of how the instruments will blend. I will be resetting all these volume settings later, but I still need a rough mix as I work.
Once all the music is done, gain staging begins
Set all volume faders to 0db so that they are neither adding nor cutting any volume.
Add an EQ to each track
Use the EQ to cut unneeded low frequencies from non bass instruments. I have a blog posts about this that might help: Video – High Pass Filter
Use EQ to cut unneeded high frequencies from non treble instruments.
Sometimes it helps to remove about 3db at 400Hz from all instruments since this is an area where many instruments contribute sound and it can all add up to a muddy sound.
Adjust the EQ volume of each track (or a trim volume can be used) so that each track peaks at about -12db (I like to use the free plugin Volume Monitor – rmsPlus for measuring this)
Apply whatever compression / effects are needed to make the track sound good
Reduce all faders to “off” (their lowest setting) and begin mixing in each track one at a time (parent faders can be used) so that it sounds good in the mix along with everything else. Opinions vary, but I like to start with the featured instrument (perhaps piano) or the most important part like vocals and set it’s volume first. Then all other volumes are set relative to that so that the featured instrument or vocal, remains the loudest part of the mix. If I want this part to be louder, I don’t turn it up, I turn everything else down.
After setting the volume faders one at a time for each instrument, this should ideally result in a mix with nothing peaking above -6db.
Now with 6db or so of space, a compressor and limiter can be added to the master bus to adjust the final volume of the song, with all tracks playing so that the loudest volume will peak just below 0db. This step is only done if there won’t be a separate mastering step. If later mastering the song, don’t add the compressor or limiter to the song, let the mastering engineer do that.
Here’s a video by Home Tracks that demonstrates the process.
Pictured below is what looks to me like a very handy chart of the frequency range of orchestral instruments. This is intentionally small because you should click the image to be taken to the interactive version of the chart.
Frequency Chart
During the mixing phase, it’s common practice to use high pass and low pass filters to remove unneeded frequency ranges from each instrument (see my blog post “Video – High Pass Filter“). For example, a flute doesn’t need to be contributing low frequencies in the bass range, so use a high pass filter to remove any low flute frequencies that are effectively noise and only serve to interfere with the sound of the bass. The opposite is also true, the bass doesn’t need to be contributing high frequencies near the top end of the flute range, so use a low pass filter on the bass. Doing this for every instrument should result in a cleaner sound because all these little things add up.
As always, use your ears when adjusting a filter, but the above chart provides a handy guide for where to place the threshold of a low or high pass filter for each instrument and also where each instrument sounds strongest.
For an example of someone explaining this chart and putting it into practice see the video below:
To me, this is a great video from Pete Whitfield showing essential ways to create realistic string parts. It covers:
Voicing of the chords – to help break away from the limits of initially performing a string part on a piano. You can only spread your fingers so far on the piano but a string section has no such restriction and can play chord voicings that you can’t play on a piano.
Sustained notes – if one a note from one chord does not change when playing the next chord (for example, playing C E G, then C F A), make it one long note. A real player would sustain the note rather than playing it twice.
Passing notes – instead of just playing one chord note and then playing the next chord note, add a note in between to lead one note to the next. For example, if you are playing a G, and the next note in your chord progression is a C above that G, perhaps add in a short B note just before the C.
Scale fragments – going beyond passing notes, consider playing part of a scale to move from one note to the next.
These are just some of the paraphrased ideas presented in the video below. This video is about the string parts themselves. As a result it leaves out one more essential tip. To make your string library sound more realistic, when moving from one sustained note to the next, ensure the midi notes overlap a little. The first note should end a little late and the next note should start a little early. A real string player allows one note to flow into the next. There is no hard break between the notes so the midi notes need to overlap.
Endless Wave – (the link leads to the Google translated page from German to English but the software itself is English) As I’ve been working to loop the individual samples in my orchestral sample library, this free tool has saved me a lot of time. Some fiddling may be necessary but with surprising frequency, it locates beginning and ending loop points that result in a seamless loop.
The idea is that you load up a waveform (you can just drag and drop a wave file) then look for a range within that waveform where the waveform looks relatively stable. While holding the Ctrl and Shift keys, drag the mouse to highlight the area where you want Endless Wave to look for good looping points and it will quickly perform a search and place a start (labelled “S” above) and end marker (labelled “E”). You can then listen to the loop. Often that’s all that’s necessary. but if you are not happy with these loop points, you can perform fine adjustments a single sample at a time, move the start or end makers to the next or previous zero crossing or simply mark out a different region and have the software search again for new start and end points.
Once you are satisfied, save the looped waveform and move on. Sometimes, depending on the variability of the waveform, it may be necessary to add a cross fade between the end and the beginning of the loop to smooth out the transition. To do that, drag the handle that is labelled “X” to the left a little or a lot depending on what sounds better. There are also 4 different types of crossover curves from which to choose. Which curve works best will depend on the waveform and the size of the crossover you’ve chosen.
Some waveforms loop very easily, others require a few tricks to make them loopable but this tool has certainly been indispensable to me.
There is further help information at the above link, and the software itself pops up a little bit of help information when you hover the mouse cursor over the various buttons and areas in the user interface.
Some tips for successful looping:
Ensure the “XFADER” button is pressed down (it’s between the two waveform boxes). This mode will display the result of the cross fade in the waveform to the left. If there is a noticeable lump in the waveform, the result will not sound good.
Try all 4 different cross fade shape buttons (middle right of the screen). When paired with the above tip, use whichever cross fade shape provides the best looking and best sounding waveform.
After using Ctrl-Shift-mouse-drag to select an area of the wave form to nominate for the loop area, drag the line labelled “x” in the large wave form display to the left to smooth out the transition between the start and end of the loop. Sometimes a small amount of distance between the “x” and “e” lines will sound good, other times a large distance will be necessary.
Sometimes a ridiculously short loop will work, provided the resulting waveform is played back with some added vibrato.
In general, playing back a looped sample with vibrato is a good way to help disguise the loop point and help the loop to sound less repetitive.
Virtual Playing Orchestra is an sfz based free virtual orchestra sample library. It’s assembled from other sfz based free sample libraries. These are some of the resources that can help with building an sfz library and understanding the sfz format..
This first link is the list of the original sfz 1.0 opcodes. These form the basis of creating a sample library for an sfz player.
There are some additional opcodes introduced in sfz 2.0 but information about sfz 2.0 is hard to find online. Here is a link to a list of the sfz 2.0 opcodes that I found:
There isn’t much of a description to accompany the list, but once sfz 1.0 is understood, it’s not too difficult to figure out what an sfz 2.0 opcode does.
That book is apparently the only place the full SFZ 2.0 specification exists. I haven’t read the book but chapter 13 and the appendix are said to address SFZ 2.0.
This next resource states “This is the main reference point for anyone who wants to create virtual musical instruments using the SFZ format.” The web site looks to me like it’s trying to be exactly what it claims. It’s definitely worth a look.
When building my library I also found this list of instrument ranges to be a helpful guide although, I’m uncertain about the accuracy of some of it. For some instruments, I’ve seen video, and heard samples, of instruments being played outside the specified range, so it’s best to double check.
Finally, although this blog post is mostly about the sfz format, my previous blog post about Looping a sample, taught me how to use the free tool Audacity to determine the values to put in the sfz “offset” opcode. Using the “offset” opcode together with the “ampeg_attack” opcode, and a few others, is the secret to creating a staccato sample from a sustain sample.
One very easy thing to do to improve the qualify of your mix is to add a high pass filter on non bass instruments.
In the presence of bass guitar, you can likely remove lower frequencies from all other instruments like acoustic guitar and even piano. Sometimes these lower frequencies from other instruments are not audible in isolation but when added together, it can muddy up the things you do want to hear in the lower frequencies like bass guitar and kick drum.
In the end, what matters is how an instrument sounds in the mix. A rhythm guitar with the low end cut a little, might not sound quite as good on it’s own, but blended with the bass guitar in the context of the mix, the song will likely sound better overall.
Here’s a short video demonstrating this from the “5 Minutes To A Better Mix Series” by “TheRecordingRevolution.com”
After reading this post: Video – Intro to Compressors, or assuming you already understand the meaning of the different controls on a compressor, the video below from “wado1942” shows a good way of setting the controls one at a time to get the desired result.
The setup before adjusting anything is as follows:
set the threshold to maximum value so that initially it has no effect
set attack and release to minimum values so the compressor starts and stops as fast as possible (compresses for the minimal amount of time)
set the ratio to it’s maximum value to maximize the amount of compression
Having set the above starting values for the compressor controls, begin to adjust them one at a time like this:
adjust the threshold to the point where the compressor is rapidly turning on and off. You want the threshold set so that the compressor is working, but not constantly on. The purpose of a compressor is to decrease the volume of the loudest parts (the peaks) of a signal, Not to decrease the volume of the entire signal.
adjust the attack and later the release, to get the desired smoothness. Do you want the compressor to suddenly start and stop working or do you want a more gradual effect?
adjust the ratio to get the desired amount of compression. I typically aim for about 3db of gain reduction.
Here’s the video that demonstrates the above steps:
This YouTube video from “The Pro Audio Files” is the best I’ve seen so far, on how a compressor and it’s settings work. Watch this video and any further discussions on the topic of compression should be far easier to understand. It clearly explains about the attack and release times, the ratio, the threshold and makeup gain. After watching the video below, you might also want to watch: Video – Setting Your Compressor
This is a 6 minute video by “Waves Audio”, focused on making MIDI brass samples sound realistic in your compositions.The most important thing to learn from this view is that, unlike string players who can play one long overlapping sustained note after another, brass players can’t do that. They need to breathe. Ensure your long notes aren’t too long and make them stop long enough for a brass player to take a breath for the next note. Using sufficient reverb on your brass samples, helps blend the long notes together in a more realistic way. Brass players sit at the back of the orchestra so in a live setting, they will experience a lot of reverb, second only to the percussion section.
The video also discusses using a couple of effects, for analog warmth and some EQ, but what I also found interesting about this video is the supporting string tracks. Notice the strings are divided up to play an ostinato pattern, some accents on the ostinato pattern and some longer held notes. A fine example of making good use of your string section instead of having every one playing the same thing or half your string section sitting silent while the 1st violins do all the work.
