Simply put---latency is the death of all recording sessions and the engineers who run them.  Comprised of a massive and simultaneous meltdown of all musicians, vocalists, and others in the session that started slowly---latency has been known to cause hearing loss from screaming vocalists who can’t seem to stay in-time as well as blunt-force trauma (and residual impact) from throwing, smashing, and splintering of guitars and other wooden, metal, and sometimes even heavy objects!  It’s also said that latency is not covered under your normal medical insurance, so proceed with caution and at your own risk!  Okay seriously---what exactly is latency, why does it exist, how does it affect me, and how can I work around/with/through/over/alongside it? 



Latency is a term that really came about as computer-based recording became more viable and affordable in the home recording and smaller project studio market.  Speaking in terms of computer-based audio interfaces, latency is the time that it takes to hear an audio signal from input to  output as you’re playing, singing, or speaking.  When recording through computer software, this is VERY IMPORTANT TO UNDERSTAND AS IT WILL AFFECT YOU!  To better understand just what latency is, imagine you’re playing that same worn-out, cheap guitar you’ve had your whole life but you’ve worked hard to finally purchase that 1956 Les Paul Goldtop and Marshall JVM215 that you’ve always wanted.  One day you called me up, ordered that guitar and amp, got it home, plugged it in, raised your arm up high, strummed the chord and you heard it come out a split-second later.....you’d scratch your head.  Now try playing several quick chords in a fast, repeating patterns---it starts to sound smeared.  If that’s not enough, now try imitating the legend Eddie Van Halen himself playing the world’s greatest solo only to find out your notes are all over the place just because of a split second delay.  In the real world, this wouldn’t be acceptable---but in the world of computer audio, it is part of life.  So what is latency in reality?  Again, it’s the delay in hearing the audio...but there is reason why this takes place.   Here’s a quick and simple test that you can do to better understand “how latency feels”.  Create a document using something like Microsoft Word, Excel, etc....  paste several heavy graphics on the page and a big text window loaded with letters, words, phrases, etc.  Then go to the end of the sentence, start clicking as fast as you can on the BACKSPACE button of the keyboard using only your middle finger and index finger.  Most likely what you’ll experience is a slight delay in between some strokes and immediate action between others.  Now imagine that’s how you listen to audio as you record---annoying, huh?


You see, computers are powerful---VERY powerful, but they were designed to do many different things.  You can’t just purchase any computer and expect to have impressive results.  In terms of Windows operating systems YES, there are things that you can do in the operating system to help streamline the computer---however you really must start from the beginning in your selection of hardware components as well as how the operating system is integrated into the computer to best reduce latency.  Even though my Creation Station computers offer the best latency performance, there are other factors that get in the way no matter how good the computer is.  Latency exists because the computer is processing the signal that you’re feeding into the audio interface, through the audio software, processing the output back through the interface, and then off to your studio monitors or headphones for you to hear.  That might not sound like much, but really it is a lot to do.  Fortunately computers today are able to handle this without much effort, but no matter how fast your computer is recording natively will always be a recipe for latency.  I mentioned “natively”, what is that?  Well, recording natively is essentially any computer-based software that does not have its own processing-based system for monitoring the recording input.  In other words if the software is using the host CPU processing (your computer’s processor) then it is considered a native recording environment. Okay, computers are quite powerful these days--but they are doing so many other tasks as it is.  There are millions of instructions going on, but which one to process at that moment is constantly in flux within a native-computer system. There are very few systems out there that have their own processing through-out (SEE BELOW), Pro Tools HD is one of them.  Everything else must use what are called “buffers” to process the audio.  The higher the buffer size, the lower the latency which means it is as close to real-time as possible.  However, this can also overload your audio system causing pops, clicks, audio drop-out or even a completely lockup or total system crash.  The higher the buffer settings, the more stable your system can function--but this makes latency even higher.  The higher the latency, the longer it takes to hear the processed audio.  And the longer it takes to hear the audio, the harder it is to truly record with.  Latency is described in milli-seconds (ms).  While this might not sound like much, you’d be surprised just how much a portion of a second really is.  Below is a chart outlining the approximate delay and affect on buffer settings.....

 

  1. Bullet   32 samples  =    .73ms delay

  2. Bullet   64 samples  =  1.45ms delay

  3. Bullet 128 samples  =  2.9ms delay

  4. Bullet 256 samples  =  5.8ms delay

  5. Bullet 512 samples  = 11.6ms delay

  6. Bullet1024 samples = 23.2ms delay

  1. Bullet   64 samples  =  .67ms delay

  2. Bullet 128 samples  =  1.35ms delay

  3. Bullet 256 samples  =  2.7ms delay

  4. Bullet 512 samples  =  5.8ms delay

  5. Bullet1024 samples =10.7ms delay

The affects of latency on an individual will ultimately depend upon their tolerance level.  In many cases, one may not truly understand just how bad latency is until they experience what a latency-free environment.  It’s said that latency isn’t an issue until it hits roughly 12ms, however many musicians will tell you anything over 5ms is noticeable.  Personally, I can feel as little as 3ms when I’m playing virtual software instruments through the computer.  You see, latency isn’t just about what you hear--it’s about what you feeeeeel.  Making music, good music, is about passion and feel....if you can’t feel it, you can’t make it.  And if you can’t make the passion, you can’t create the feel for others to listen to.  So yes, latency is an important issue.  The reality is that if it’s just one person and their home studio, they usually find a way to deal with it---however if you’re recording other people for money, latency can literally drive paying customers away.  When you have customers paying to record, they shouldn’t have to change how they play to work in a particular studio.  Remember, the more audio channels that have to be processed at a time--the more that piles up.  And the more latency that piles up, the worse it gets.  The numbers you see above are essentially best-case scenarios BEFORE you insert software plugins, create monitor and aux busses, and route through hardware inserts.  Those things can easily double, triple, or even quadruple the delay making it nearly impossible!  Obviously, that’s just not acceptable to anyone!

Okay, I mentioned a system called Pro Tools HD that has its own processing.  The short definition is that this system has dedicated PCI cards that plug inside the computer with its own processing chips that handle ALL aspects of the audio including input and output monitoring, processing of the plugins right on the hardware, ADC (Automatic Delay Compensation) to keep everything in perfect sync, processing of the audio, writing to disk, etc.  Most other systems rely on the computer’s processors (as mentioned above) and have to share the processing with anything and everything else the computer must handle, this (again) is known as NATIVE PROCESSING.  This can slow the computer up meaning, you may not have as much processing as you thought depending on how complex your mix gets, how many plugins you’re running, how complex those plugins are, etc.  The more intensive this gets, the worse the latency can be.  This is why the dedicated DSP processing provides predictable results and reduces latency.  NOW THAT WE’RE ON THE SAME WAVE-LENGTH HERE....... There are many audio interfaces that have DSP monitoring built-in allowing for latency-free monitoring, but there is a catch!  You see, these interfaces do exactly what they claim to do and that is to provide this processing on the interface itself relieving the stress on the computer to monitor your input.  HOWEVER should you monitor through plugins in your software, then latency will be introduced.  Why is that?  Well, you’re now monitoring right back through the computer’s processor instead.  This is the same for products like the Universal Audio UAD2  cards or the TC Electronics Powercore.  These are amazing tools for mixing providing the quality of high-powered DSP-based plugins for your audio applications, but they are meant for mixing---NOT TRACKING.  Although there is DSP processing, that is for powering the plugins---the reality is that SAMPLE BUFFERS are still needed to process the plugins through the audio environment which is back to using the native computer processor.  Make sense?  Now, some interfaces do provide built in effects processing for monitoring the input on the hardware thus eliminating latency when recording---very cool!!  Here are a few examples:

MOTU

Firestudio TUBE

Dealing with latency is part of the game when recording through a computer system, but there are ways to deal with it if it gets to be too much for you or your clients to deal with.  There are many interfaces, such as the ones listed above, that offer onboard processing to enable the user(s) to listen to the audio in real-time without latency---that’s the good news. There are also some that have built-in effects processing allowing for the real-time use of reverbs, delays, and even compression or EQ without added latency.  But if you are using a product that does not support the above, there are ways to help improve the latency--depending how well-tuned your computer is, however.  Make sure that you have your sample buffers set as low as possible without popping, crackling, or causing your computer to become un-stable.  Don’t monitor with high-powered plugins such as the Waves Studio Classics or Altiverb from Audio Ease.  Some plugins are not that processor intensive, so seek those out by the old-fashioned “trial and error”. 


Other options are to just monitor the way that it used to be done---THROUGH A CONSOLE.  By placing a mixer on the front-end of your audio system, you’re now monitoring without worrying about the computer processing the information.  This also provides a quick and efficient way to create headphone mixes, eliminate latency altogether, and improve workflow as well.  If you choose this direction, make absolutely certain that the mixer is of good quality and has direct outputs for recording on EVERY channel.  If you compromise on either of those points, then you will do more harm than good.


Obviously this is only scratching the surface here---if you have any questions, please do not hesitate to contact me via email or by calling toll-free at (800) 222-4700 ext 1362.  I’m a professional audio engineer that provides real world solutions for real world needs.

 
There are basically (3) connection possibilities for an audio interface being USB, FIREWIRE, and the traditional PCI slot within a tower computer. But within each protocol there are different versions. For example USB currently is known as either USB 1.1 or USB 2.0. The difference in performance is quite significant with USB 2.0 providing better data streaming, lower latencies, and longer thru-put over time as compared to the previous USB 1.1. Most interfaces these days 2009 and later are USB 2.0 compliant, but you have to make certain that the USB ports on your computer are also supporting USB 2.0. Fortunately, that too is also pretty standard these days. In fact, you have to look pretty hard to find a new computer today that is NOT USB 2.0 compliant. When comparing to Firewire, there are some significant differences in performance. Also known as IEEE 1394 (IEEE 1394a = Firewire 400, IEEE 1394b = Firewire 800, Firewire also comes in two distinct forms known as Firewire 400 and Firewire 800 (which I’ll refer to as FW400 and FW800). Essentially, FW800 has twice the thru-put of FW400, however FW400 is nothing to shake a stick at! On paper, however, USB 2.0 appears to have a faster bus thru-put over FW400 with USB 2.0 offering 480mbps over FW400’s 400mbps (and of course, FW800 = 800mbps).....BUT NOT SO FAST.....This is on paper, but in the real world FW400 trumps USB 2.0 in many ways. In terms of data thru-put such as spooling multiple audio files or streaming software samples, long videos, multiple video clips, etc from hard-drives--USB varies its transmission speed depending upon the data its reading or writing. USB also does not have the same burst rate (start) that FW400/800 has, it technically must “ramp up” therefore it’s not a good performance standard for we audio and video type people. THIS IS WHY I STAND ON THE SOAP BOX AND NEVER RECOMMEND A USB HARD-DRIVE FOR ANYTHING AUDIO or VIDEO RELATED! This is also why I say, BUY GLYPH DRIVES---there are many reasons, to find out why CLICK HERE. When comparing audio interfaces, USB 2.0 also doesn’t carry the same performance of its firewire counterparts in terms of monitoring. There is an inherent latency that is built into USB, more-so with USB 1.1. This is why you find very few audio interfaces that have more than (4) inputs that are not firewire. Another thing is that there is less power sent through the USB channel as opposed to firewire, so the performance is generally less in terms of microphone pre-amps, gain-stage, signal to noise, etc. Firewire certainly has quite a bit of bandwidth, enough for most people and users. But there are those of us that need more than 32-inputs (the general safe-range for audio channel thru-put)...this is where the traditional PCI slots come into play. I’m not going to focus on this but some products include the MOTU PCI-424 series, Digidesign’s Pro Tools HD, the RME HDSPe series of ADAT I/O cards, Apogee’s Symphony64, etc. Why_Glyph_Drives.htmlhttp://www.theaudioprofessional.com/MOTU/PCI-424_Card.htmlshapeimage_15_link_0shapeimage_15_link_1shapeimage_15_link_2
 
All information, images, and content on this website are copyright by Brad Lyons and www.theaudioprofessional.com All Rights Reserved. If you have any questions for Brad Lyons, please call (800)222-4700 x1362 or email directly. http://www.theaudioprofessional.commailto:questions@theaudioprofessional.com?subject=QUESTIONSshapeimage_16_link_0shapeimage_16_link_1