Welcome to Barnjazz Riffs

I finally decided to quit wimping out and start submitting my music to Film and TV opportunities. I joined Broadjam as a primo member so submitting is pretty cheap, anywhere from $2.50 to $5.00 per tune. Since I do mostly instrumentals its sometimes hard to find good fits, but recently there have been alot of requests for new age, acoustic, and other styles that I do. I also submitted to a dance/techno category for the tune “Black Feather”. Not too bad for a boomer, right! You can listen to the tune by going to the Broadjam site and searching for artist Diamond Jim.

So its the old “you can’t win if you don’t play” routine. Kind of like the lottery, since hundreds of musicians are submitting original songs and tunes to these opportunities. Wish me luck, dear reader!

When mixing and mastering, there are many ways to approach the workflow, especially when working both inside and outside the “box”. In this case my “box” is an Intel Quad Core, 4 Gig, XP Pro system using SONAR 8.5 PE for tracking and mixing (and occasionally Reaper 3), and Wavelab 6 for mastering and CD production.

Here is my workflow:

Using SONAR, where I originally laid down the tracks, I mix out through my console and outboard and then back into SONAR, and export the final stereo mixdown to a project \Mix folder. The mix version retains the original bit depth and sample rate, usually 24/48 or 24/88.

In Wavelab I load each mix file separately for mastering, create the master section fx chain, and save that as a per-song preset. I will set the loudness using the WL meters, or the meters in UAD Precision Limiter, set to K-14. I also use UAD Precision Maximizer, or possibly the Fairchild 670 plug, to get the sound where I want it in terms of loudness and clarity, warmth, etc.

Then I render out to same bit depth, with final gain set for the target avg RMS (I use the Global Analysis tool all the time to check this), generally -14 DBFS plus or minus a db. No dither yet. Note that this loudness level is average perceived loudness, set to less than today’s pop music standard, i.e. it retains much of the original dynamic range and is still loud enough to sound good on CD compared to many other similar tunes. As I am mostly doing acoustic, folk, Celtic, electronica, country and bluegrass, i don’t feel the call to squashing the sound just to make it competitive for radio play, as has been the trend in recent years. Anyway…..

Each of these 24-bit mastered files is kept in a \Master folder in the project subdirectory, so I can keep track of when I last rendered out a final version. At any time I can go back to the mix versions restore the preset for the tune, and work it some more if I need to.

I create a CD project montage in Wavelab, set order of the tunes, do the fades (top and tail), spacing, and only then apply dithering to the final rendered stereo file and basic audio CD file. That way I can adjust final levels in the montage if I want to before dither.

It might sound like a lot of work but it gives me pretty good control over each song, without having to render the entire CD montage with all of the original per song plugs, which takes a while. The final render only applies dither.

After the final render of the project I can then burn reference CD’s, which i listen to in a variety of contexts (home stereo, car stereo, iPod). Once its all good, a final master is burned onto a high quality CD-ROM, such as Apogee or Tayo Yuden, at 8X speed (which my DAW seems to like better than the often quoted 1 or 2X speed).

Here are the clips I referred to in my previous post about the effects of partial phase cancellation when recording vocals with two mics and two artists. The first clip shows the results of two tracks, each with the vocal mic but the bleed from the second mic is retained. The second clip shows what happens when I edited out the bleed on the second mic track.

PhaseExample_In

PhaseExample_Out

The Setup

Just finished a voice over recording session for a radio play. The play will be on an upcoming CD release for the heavy metal group, Butcher. The voice over was performed by two actors spaced 3 feet apart, facing each other. The large diaphram condensor mics were set to a cardioid pattern, but of course there was considerable bleed between the two.

I used a U87a on the female actor’s voice, and a Michael Joly modded MXL V67 on the male’s voice, each recorded to separate mono tracks in SONAR. The U87 was clear and powerful, full of presence, as I expected, but the sound from the V67 was thin and hollow sounding, which I noticed only when playing both tracks back together. This was not what I was expecting from the Joly modified V67.

Then it hit me! The bleed into the U87, being 3 feet away, was introducing a 3 ms partial phase cancellation for the male actors voice, when listening to the U87 track.

Its about 1 millisecond delay per foot of separation, that’s recording 101 for you. And yes it was considerably noticeable. However when I muted the U87 track and listened to the male actor voice soloed, it sound very good. Not U87 good, mind you, but much better. Considering the V67 is a $300 mic (after the Joly mods), it holds up well as a 2nd stringer to the U87.

So rather than try a different mic I just deal with the phase cancellation in the DAW, when mixing. Its a simple matter to edit the tracks to eliminate the bleed, as long as the two voices are reading distinct parts and not talking over each other. Fortunately that is the case for this script, just a few places where you will hear both voices together.

Being out of phase is something that recording engineers really need to pay attention two, especially for vocals, when recording with multiple mics. Having better isolation also helps, but I did not have the room nor the available “gobos”. I might have used one of Ethan Winer’s Real Traps Portable Vocal Booths in this situation, but I wanted the talent to face each other as they were carrying on a “conversation”.

I will post some examples of this partial phase cancellation for anyone to hear, so you know what I mean.

Soundtracs PC Midi Console

So on to calibrating the console. This was more of an attempt to see if there were any level differences between any of the 16 channels. I have an older Soundtracs PC Midi 16, which was originally manufactured in Great Britain in the 80’s. This is a large, heavy beast for just having 16 channels, but I like the sound I get. However being an older console its important to check to see if any of the capacitors (caps) are wearing out, if there are wiring problems, etc. I have not opened it up to inspect it (yet), so for now I am just going to run signal into it and see if I can hear any major problems with any of the channels.

Since I record both ITB (In the Box) and OTB, using the console, I have spent some time rigging a series of patch bays which will allow me to easily switch between recording to tape and to the DAW, as well to configure how the output from the DAW is routed. Normally I route a stereo pair of outputs from the Lynx Aurora convertors to my Central Stations monitor controller.

When I want to use the console for analog summing, as well as to take advantage of mixing using real faders, EQ, etc., I patch all 16 channels out of the Lynx into each of the 16 Tape Inputs of the console, using the patch bay. I have rigged a pair of 8-channel snakes, one set for the outputs, and one for the inputs, into the patch bay, so switching my routing patches is pretty easy and fast.

Patch Bays

So I set up the console to take in all 16 channels from the Lynx, and played the Pink Noise signal through each.
The levels on all but channels 15 and 16 matched perfectly, showing about +3 dB on the LED meters. I then adjusted the master faders until the stereo output was set to show 0 dB on the master output meter LED. This is not exact, using these LED’s but its close enough so that now I have a good reference level. The setting of the master faders is about -6 dB to attain 0 on the output. I have not actually measured any electrical signal levels, yet, which is what I would need to do to get a better calibration.

I did find something interesting in channels 15 and 16. They were considerably hotter than 1-14, by about 10 dB. This led me at first to conclude that something bad had happened to the channel electronics. However, the difference was the same for both channels, not a likely occurrence. So I patched the 16 input into 14, and, yep the same level difference was noted on the meters. So this was coming from the DAW!

So fired up the Lynx Aurora mixer software, and noticed nothing amiss in the soft fader settings, they were all set at 0 (max). Now this software mixer is not the most intuitive piece of soft kit, so rather than try to figure out why the last two channels were hot, I hit the software factory reset, and that fixed it! The levels are now the same across all of the 16 channels of the console.

So the lesson is, don’t blame the old guy. Sometimes you have to spank the baby. Ok, bad analogy. But don’t blame the hardware, which could lead you down a path of big expense and waste of time, before you check to make sure your output from the computer is set correctly.

It’s amazing what you can learn about the short-comings of your studio setup when you take the extra step of calibrating your monitors and console. After a long several months of integrating an analog console (Soundtracs PC-MIDI 16) into my DAW setup, including installation of patch bays, cabling, and new audio interface (Lynx Aurora 16), and some nice used Genelec 8040a monitors, I finally got around to running some fairly simple and standard calibration tests.

So what is calibration for anyway?
When you are mixing an audio project you make decisions based on the frequencies and loudness of various tracks. Due to psychoacoustic factors (how the brain perceives audio) we are often led to make decisions based on loudness (louder often sounds better) that don’t always contribute to the overall quality of a mix. Many audio engineers agree that listening to mix at high loudness levels is not only bad for your ears, but may not result in a balanced, pleasing mix.

Bob Katz, the renowned mastering engineer (www.digido.com) has advocated the K-System of metering, and this involves among other things having a well-calibrated audio system where the maximum loudness of your monitoring system is set to a known value.   Here is what I did last night to get my loudness levels set correctly.

1. Downloaded the -20 DB pink noise wave file from his site (http://www.digido.com/index.php?option=com_phocadownload&view=category&id=1&Itemid=83).
2. Turned the volume control on my Central Station monitor controller all the way to the right (to zero db), and turned down the trim controls for my main monitors so I would not damage my speakers and ears.
3. Turned off the right speaker, so I would be calibrating one at a time. 
4. Loaded the pink noise file into Wavelab (any DAW would do), and set it to loop continuously.
5. Pulled out my trusty Radio Shack digital SPL meter, set it to C-weighting and slow response (per Mr. Katz’s recommendations), and sitting in the listening position, pointed the SPL meter at the left speaker.
6. Adjusted the trim on the monitor left channel louder and louder until SPL read +83 dB. This becomes the calibrated listening level for maximum loudness, when my monitor controller is set to zero dB.
7. Turned off the left monitor, turned on the right one, and repeated steps 5 and 6.

The Results, Please!

OK, this was pretty loud, although pink noise is a random noise,  and 83 dB is quite tolerable. But I neglected to do one thing.  I forgot to disconnect my subwoofer!  OK, I have a KRK sub patched in line with the main monitors, and realized that the considerable amount of bass noise I was hearing came of course from the sub, which sits on the floor underneath my workstation. 

So I disconnected the sub and hooked the output from the Central Station Monitor A directly into the Genelecs (there is no way to otherwise bypass the sub, unfortunately). Then I reran the calibration again and found that my SPL’s were down considerably, to about 72 dB max. The subwoofer was adding a fair amount of low frequency audio to the mix.  This may or may not be a good thing, depending on what you are mixing, but I decided that it would be better to start clean, without the sub. 

So I recalibrated and reset the trim levels on the CS so that now I am back to 83 dB for each monitor. Life is good.  Loaded a mix I had been working on, and began to immediately notice some ickiness in some of the synth bass lines, violin etc.  I had been using plug-in EQ to play with this sound, and now it sounded pretty bad.  So I disabled the plug-ins, and basically decided to remix the project from scratch.  I noticed that the Genelecs now yielded a more “accurate” mix.  I did not need to monitor at 83 dB, I actually preferred setting it the level about 10 dB lower, but when I want to hear it louder, I can now be reassured that my level won’t exceed 83 dB.  This will hopefully add more consistency to my mixes, as well as protect my ears…

Next up - calibrating the console.

So I was laying down a violin part to a new tune last night, trying to follow a synth piano line that I had recorded several months ago. I was also trying out my new Neumann microphone to see how it sounded with the fiddle.

Its amazing how much worse my intonation was when I compared it to the synth piano part. It was not pretty!

But then when practicing alone, without a partner (synthetic or real), I think we tend hear what we want it to sound like. We don’t notice the out of tune notes until they are compared against a perfect standard.

One thing, be sure your instrument is in perfect tune before trying this. I compared my out-of-tune notes using Melodyne, and notice that I was consistently sharp most of the time rather than flat. That indicates to me that either I am playing sharp tonight, or else I am tuned up 5-10 cents. 

The other thing, is sometimes you are just not spot on.  I was thinking the whole time, that a classical violinist would not have this problem. But I know better.  Maybe a trained professional, who practices 6 hours a day, would have less of a problem, but we all experience those times when we just can’t play in tune, to varying degrees.

The final thing, I did find that by the 5th or 6th take that my notes were much closer to being in tune.  I was finally able to get a take that I did not have to “repair” using Melodyne.  I also realized that my headphone mix was not very good, that the audio I was playing along with was too loud compared to my instrument mix. So I am going to work on getting a better headphone mix.  I am certain that if I had a client/musician in the studio that they would perform better with a well-balanced mix.  In this case, I was the client.

So yes, practice does make perfect, or at least better than “it sucks”.  But there are times when you will suck, and others not so much. As musicians we have to keep reminding ourselves of that. We are not synths.

I recently gave a talk to the San Manuel Rotary breakfast club where I was allowed to expound on the pros and cons of digital versus analog recording to a group of hungry business persons. Thanks for putting up with me, Rotarian folks, you are great people!

So here is the gist. In the beginning of recording there were wax cylinders, then wire (magnetic recording on actual lengths of metal wire), then finally after World War II, tape. We can thank Studer, the German company that invented tape recording for Hitler, for the format that dominated recording up until the 1980’s. Studer tape machines are still in use in studios around the world, and are considered a kind of gold standard for tape decks.

OK, that brings us to 1982, when the compact disc format was unleashed upon the world. Up until then, analog recording from the tape machines went directly to vinyl. Many of us cherish our collections of old LP’s for their warmth and smooth sound. With the CD came digital recording. Digital is a format where analog sound is converted to numbers, sampled at rates of 44,100 times per second (abbreviated as 44.1 KHz) and higher. At first CD’s were hailed as a revolutionary advance, but soon after the audiophile community began to notice that CD recordings often sounded more sterile or brittle compared to vinyl.

Now the early CD players had rather primitive converters compared to today. A digital-to-analog converter is a chip built into the CD player which translates the numbers back into audio waveforms. Similarly an analog-to-digital converter was used in the original mastering of the CD to get the analog sound from the tape into the computer for burning a CD. Today the converters are not an issue, except in low-end CD and DVD players. 

The “Red Book” CD format specifies a much lower resolution (16-bit) and sample rate limited to 44.1 KHz, compared to how music is actually recorded digitally. Most engineers record at 24 bits or higher (more bits means more headroom, that is, greater dynamic range between the softest and loudest sounds). Sample rates of 96 KHz or higher are not uncommon, although there is a raging debate about how important these are.  To make a CD we have to reduce the bit depth and modify the sample rate to match the CD format (this involves the technical processes of dither, and sample rate conversion, respectively).

The biggest problem with digital audio today is not that it is inferior, but that it is too good!  It really does record what is out there with astounding clarity. However that clarity includes all the warts of a poorly treated recording room, irritating resonances of some instruments (such as the violin!), etc.

So what does this have to do with a digital-analog hybrid studio?

Sometimes we would rather view life through rose-colored glasses. With audio we often prefer the more “rounded” warm sound of analog tape/console circuitry (which actually adds some pleasing harmonic distortion), to the often clinical sound of digital.  That being said, there are ways of making digital sound better, and I have explored many of these. Aside from software plug-ins, which attempt to simulate analog warmth, there are other approaches which take advantage of real analog hardware.

One trend in recent years has been to go all-digital, recording directly from a microphone or instrument into a pre-amplifier, and then into the computer via a “digital audio interface”, basically an A-D convertor hooked up to the computer via Firewire, USB, PCI card, etc. This is assuming that the musician/engineer is recording any kind of live instrument or voice to begin with. Much of today’s electronic dance music is made entirely “In the box” using plug-ins and software synthesizers running in a piece of software often called a DAW (Digital Audio Workstation), such as Protools, SONAR, Ableton Live, Cubase, Logic, etc. (I prefer to use the term DAW to refer to both the software and the computer hardware).

But – A counter-revolution is underway, fueled by a conspiracy of old school guys and young engineers who record bands or themselves playing real, actual hand-on musical instruments. This approach acknowledges the advantages of digital audio for editing, applying signal processing (compression, EQ, pitch correction, etc.) while at the same time trying to get as much of that old-school analog sound as possible.

There are many ways to approach this:

1. Record into the DAW, then send your final stereo mix out to analog processors for sweetening, then convert it back into digital for the final mix.  Devices are available called summing boxes that allow you to take the converted audio, run it through a tube pre-amp, compressor, eq device, etc. before routing it back into the computer.
2. Record using a console, then into the computer. Consoles were mandatory in the older days, but now most small, project or bedroom studios cannot afford them, or even have enough space for one. Basically a console (or “desk” as it is often called) is a series of channels with faders and knobs, preamps, EQ, routing capabilities, etc. with as few as 8 and as many as 100’s of channels, each capable of recording an instrument or vocal simultaneously. Some consoles are entirely analog, some are digital, and some both. Old school analog consoles can often be found used for a “song” (I paid $500 for mine, it cost somewhere around $8000 in the 1980’s). Use the high quality analog circuitry of the console as front-end to your DAW to get some of that sound we strive for.
3. Even more radical, get a used tape machine, record through the console to tape, and then dump the tape tracks into the DAW for final editing and processing.
4.  Any combination of the above, where you involve real analog hardware. For example some engineers advocate taking digitally recorded mixes, sending each track or group of tracks out the D-A converters into separate channels of the console, then letting the console produce the final stereo mix (“analog summing”), then sending that back into the computer. This is similar to #1 but involves the sound introduced by the analog circuitry of the console, where the console electrically “sums” or combines the separate tracks into a stereo mix. 
5. You can also take stereo or even individual tracks that were recorded entirely in digital, and send them out the tape machine and then back into the computer. This uses the  tape machine as a kind of effect processor.  Tape is known for its relative warmth, a byproduct of way that tape stores electrical information. That is topic for another blog entry, however.

Any or all of these approaches can be found in the modern-old school digital-analog hybrid studio.  I am sure having fun implementing mine.  Everything old is new again…..

You can leave a comment by clicking on the Comments link at the bottom of each post. If you are not logged in and have not done so, you will need to register a user name and email address. More info can be found in the About link at the top right.

I had nothing else better to do the other day so I decided that we need a version of the MD’s Hippocratic Oath for recording engineers. Why not?  So I borrowed the language from a modern version of the MD’s oath from here.

The Hippocratic Oath For The
Recording Engineer