How Good Can It Sound?

Stereophile magazine has published an article on iTunes. Most of it is yet another rehash of new features of iTunes 4.5. However the most interesting point is that the New York branch of the Audio Engineering Society (AES) is gathering this month to discuss the impact of portable digital audio players like the iPod using compressed audio files on the sound of music.

It is an important topic. When we do the final mix for films, we go to a dub stage that looks like a theater with most of the seats pulled out and an enormous mix board in the middle of the room. The speakers on these stages sound far better than the ones in most movie theaters. When it comes time to do the DVD mix, sometimes near field speakers are placed on the stage. These might be Genelecs or Audix or some other high end speaker that sounds much better than your home theater setup. The point is we try to get the best sounding mix out of the best sounding speakers. And when we go from movie theater to home theater we try to accurately reproduce the ideal listening environment.

Another way to think of it is that you know a hundred dollar pair of headphones is going to sound a lot better than a ten dollar pair. When you’re working with sound, whether it’s music or films, you want it sound the best it can under the best possible situation. (And therefore it should sound as good as possible under less than ideal situations.)

But iTunes and iPods and the like present a different problem. We are no longer just talking about monitoring through different headphones or speakers. We are talking about reducing the quality of the sound before it even gets played back. By it’s very nature, a compressed sound file is better at playing back certain frequencies, and worse at others. In my own personal observations, high frequency transients like the harmonics from cymbals are the first things to get thrown out in MP3s. So if you know that a lot of your listeners are going to be listening to MP3s versus CDs, do you start reducing those high end frequencies in your mix?

It’s definitely something to think about and should be an interesting discussion.

One Step Beyond

A Pro Tools update for everyone.

I’ve said that we set up a Pro Tools system in an unsupported configuration to and have been testing it. You can get all the details on the system in my earlier post.

It works pretty darn well. We had our first big test Thursday and Friday with pulling sound effects from our new Soundminer database and sending them into Pro Tools to cut a very action intensive 5 minute scene. Guns, explosions, general mayhem. It worked beautifully. I don’t remember the exact track count, but it was over 32 because at one point Cameron had to change the voice setting up to 64.

The one disappointing thing was the performance of the MJPEG A quicktime movie on the Aurora Igniter video card. It was a bit jerky. We checked the Info window while playing the movie in Quicktime Player and it wouldn’t stay at a constant 24 fps. Occasionally it would speed up or slow down by about .5 fps. We didn’t have time to tweak the settings so I can hopefully get that to perform better.

Plus it wasn’t exactly the standard Quicktime file that we would normally playback for sound editorial. It wasn’t loaded off a video tape like they normally are. This was the conversion from an Avid Quicktime that I mention in my last post. The image size was larger than I normally digitize at, and maybe that had something to do with the less the perfect playback.

In fact, the Avid Quicktime played back better even in Pro Tools with all those tracks of audio running too. The only downside to that was the movie could only display on the computer monitor and not on the video monitor. So Cameron used that to cut against instead of the MJPEG A picture.

We haven’t really tried out the SCSI on this setup yet. This scene was cut off the internal hard drive (a standard HFS+ format, not journaled) and the digital picture was played back off a second internal hard drive.

It’s not the be-all end-all, but it does seem that the Pro Tools 6.2.3 software works well with Mix hardware on a G4.

Avid Quicktime Codecs

This one is for me as much as anyone else.

Once every six months or so I find myself in a situation where I need to either play an Avid Quicktime or convert it to another codec that’s more useful to me (like MJPEG A), and I’m on a system that doesn’t have the Avid codecs installed.

Yesterday I had to do it again. And I had a hard time finding the right codecs on Avid’s website again. So here they are in all their glory:

Avid Meridien Quicktime Codecs for Mac OS X, Mac OS 9, Windows XP, and Windows 2000. These are the latest versions.

Avid Quicktime Codecs for Media Composer / Film Composer 10.5 and Xpress 4.5 (Mac OS 9). This includes ABVB 9.3 and Meridien 9.4.1.

Avid Quicktime Codecs for Media Composer / Film Composter 10.1 and Xpress 4.1 (Mac OS 9).

Avid Quicktime Codec ABVB 8.0.2 for Mac OS.

Avid’s software download page. Though for some reason you can’t actually find the latest Meridien drivers on this page.

Don’t forget to put your Quicktime codecs in the right place! In OS X, they need to go in /Library/Quicktime. In OS 9, they need to go in System Folder/Extensions. (Windows users, I have no idea. Sorry.)

Codecs for OS 9 will not work in Classic under OS X!

Lossless Is Good

If you keep up with Apple developments at all you probably already know that today is the one year anniversary of the iTunes Music Store. In honor of that Apple released iTunes 4.5 and Quicktime 6.5.1.

The new version of iTunes includes features like the Party Shuffle, Jewel Case Insert Printing, and iMix. You can read about all that stuff over at Apple’s site or pretty much any other Mac news website.

The feature that I found most intriguing is the new Apple Lossless Audio Codec that comes with Quicktime. MP3s and AACs have revolutionized how we think about music. But let’s face it, they are compressed audio files. Any CD will sound better. iTunes and the iPod will quite happily play uncompressed AIFF files but they are much larger. You are probably familiar with an MP3 at 128 kbps. An uncompressed AIFF file runs at 1411 kbps. In other words, 0.94 MB per minute versus 10.33 MB per minute.

So I pulled out my CD of Jet’s “Get Born” and decided to do a little experimenting with “Are You Gonna Be My Girl.” Marc Heijligers did some excellent qualitative studies of various types of audio compression in “Encoding Observations.” I used a method similar to his to analyze Apple’s Lossless compression.

I used Peak to rip an uncompressed AIFF of the track. I opened that file with Quicktime Player Pro and exported it to a movie with the Lossless codec. (Unless I missed something, it seems that only the “export to Quicktime Movie” option allows you to use that codec.)

Export to Quicktime Movie with Apple Lossless Compressor

Just to be completely legit, I exported another movie with the audio uncompressed. The first thing I noticed was the difference in file size. At 3:33 long, “Are You Gonna Be My Girl” was 36.1 MB uncompressed and 27.1 MB with the lossless compression. The compressed file was 25% smaller. Again in comparison, a 128 kbps MP3 is 91% smaller. For this song it would be about 3.3 MB

Obviously an MP3 or AAC is a lot smaller, but you also lose some of the sound information in the process. This new Apple Lossless codec promises to save 25% in file size and is supposed to sound exactly the same. To test this out I opened both the uncompressed movie and the lossless movie in Peak.

Visually the two files seemed to have identical waveforms.

Uncompressed Audio
Uncompressed Audio

Lossless Compressed Audio
Lossless Compressed Audio

With a quick listen on my studio headphones, I couldn’t tell the difference. I needed to be sure though that there was literally no difference. This is where I used a method that Marc talked about in his article.

I inverted the phase of the Lossless compression file.

Invert Phase

Then I copied the entire file to the clipboard, and using Peak’s “Add” DSP function, I mixed the phase inverted Lossless audio with the original uncompressed audio.

Add Function

The resulting audio file was completely silent.

Silent Audio

A sound waveform looks a squiggly line drawn along the X-axis of an X/Y graph. You might be familiar with a Sine wave from trigonometry. A simple tone looks like that. When a sound is phase inverted, the peaks and valleys of that squiggle are swapped. So where the original sound might have a peak at 4 on the Y-axis, the inverted sound would have a valley at the same point in time at -4 on the Y-axis. Obviously if you add 4 and -4 you get zero. So a phase inverted sound mixed into the original sound should give you a silent audio file.

Since I phase inverted the compressed file, mixed it in with the uncompressed file, and wound up with a silent file, the codec truly is lossless. The two files are sonically identical.

Out of curiosity, I imported the original AIFF file into iTunes. I went into the preferences and changed my import settings to “Apple Lossless Encoder.”

iTunes Importing

Then I used the “Convert” function under the Advanced menu to convert my AIFF into a compressed file using the new Lossless codec.

Convert Selection

The resulting file was an AAC file with the same .m4a extension as my other AACs. When I got info on the new AAC, I found that it was 1061 kbps.

Info for Uncompressed Audio
Uncompressed Audio

Info for Lossless Compressed Audio
Lossless Compressed Audio

With a space savings of 25% and truly no loss of sound quality, Apple’s new codec is definitely something to take a look at for both sound professionals and audiophile consumers.

Digital Audio Field Recorders

For many years the DAT recorder has been standard for both timecode and non-timecode field recordings. Now that Digital Audio Workstations are supporting file standards beyond the 16 bit, 48 KHz limit of DATs, it makes sense to look at other options. I was able to check out 3 cool new recorders at NAB.

Deva V

Zaxcom’s Deva recorder is in essence the grandfather of the hard disk field recorders. There have been others but they are the first to really offer one that has gained widespread use in production sound.

Their latest recorder, the Deva V, which is due “any day now”, is a big improvement over the 4-track, 24 bit, 48 KHz Deva II. The new recorder offers 10 tracks of audio at up to 192 KHz. It also has a touch screen menu which offers much better access to functionality than the few buttons of the Deva II.

In the demo I saw, the Deva V has an on-screen keyboard which allows you to enter text like track name information, and scene and take, which was only previously possible through an external digital mixer with a built-in keyboard. The Deva V also allows you to route any input channel to any record track (or number of input channels to number of record tracks for internal mix-downs). The older Deva II had a similar functionality but the new track matrix screen makes it much easier to see what’s going on. (Ask any Deva II user about their first time with the machine and they will probably admit to multing Input 1 to every record track for a while. It was a common mistake.) The only problem with these on-screen functions is that the screen is quite small so you need a pointer similar to a PDA pen to select many of these functions.

They’ve also added a Firewire port to the device and and optional internal DVD-R. The DVD-R is a smart move away from DVD-RAM. If you’ve read my previous article on the headaches of DVD-RAMs in the Mac, or you’ve experienced it yourself, you know what I’m talking about.

It’s a cool device and I’ve only just touched on some of the new features. It is a bit pricey though. It will set you back about $13,000.

Sound Devices 722 and 744T

These two recorders are very impressive. They are extremely small–a little larger than a VHS tape. They have a very clean and simple interface. According to the reps I talked to at NAB, they are 60 days away from release.

The 722 is a 2-channel recorder, while the 744T is a 4-channel timecode recorder. They both do 24 bit recordings at up to 96 KHz. The 722 has an internal 20 GB hard drive which would give you about 30 hours of stereo recordings at 16 bit, 48 KHz, or about 10 hours at 24 bit, 96 KHz. The 744T has a 40 GB hard drive, so it would give you about the same amount of time but with 4-channel recordings. (Or double the times of stereo.)

It has a Firewire interface, so that when you plug it into your computer, it shows up as a hard drive on your desktop for transferring files. It also has what they call a C-Link interface which allows you to daisy chain multiple recorders together and have them all going into record at the same time with the push of one button.

The price is right on these as well. The 722 will go for about $2600 and the 744T for about $4200.

Fostex FR-2

This is the recorder that impressed me most. It’s about twice the size of the Sound Devices recorders but it is very light. Like the others it records 16 or 24 bit sound files, but it can go all the way up to 192 KHz. It’s only a 2-channel recorder, but there will be a timecode option available later this year. It is also in stores right now.

It only has a USB interface to attach it to your computer and transfer files, so it is not as fast as the FIrewire of the other recorders. However, the $1300 price tag more than makes up for a little extra transfer time.

The one caveat is that it doesn’t have a built-in hard drive. You have to purchase that separately. This is can be seen as a bonus though too. It supports recording to 1.8″ Type II PC Card hard drives. Many of the 5 GB models now go for about $250. The other option is that it supports Compact Flash Type II cards for recording. The same ones used in some digital cameras. The 2 GB cards go for about $400 right now. It is an additional expense, but there are no moving parts. Plus by recording to compact flash, you can pick up a $20 USB compact flash reader and transfer your recordings to your computer without eating up your recorder’s battery time. With a 2 GB card, you can get about 3 hours of 16 bit, 48 KHz stereo recordings or 1 hour at 24 bit, 96 KHz.

The Quest For The Knob

t goes without saying that the most important thing about an audio workstation is that you are able to hear what you are working on. This often leads to the continual quest for the best sounding speakers and headphones. The question that those of us who work in the sound field don’t always think about is “What is the best mixer to use?”

A standard Pro Tools system has 8 outputs. (Yes, the HD hardware can output up to 16 but often on a standard editorial system, people only use 8.) The various flavors of the Mackie 1604 has been the work-horse of mixer setup.

However, ever since Pro Tool 5.1, the software can be configured to handle an internal surround mix and output in 5.1. With this change editors using workstations that are configured to monitor in 5.1 really only need to use 6 outputs. So on the mixer side of things, you need the 6 or 8 inputs from Pro Tools and probably another 2 (stereo pair) for the Mac speaker. Sometimes you might have other gear like a DAT or video deck that you want to monitor independent of Pro Tools, but often you just need the 8 or 10 mixer inputs to handle everyting.

The problem comes with the outputs to the speakers. The easiest way to deal with a 5.1 speaker setup is to assign each bus out to a speaker. So you need 6 buses to handle the outputs. The Mackie 1604 only has 4. You can make use of the Aux Sends for the two additional outputs but it’s not configured as nicely. So now you’re looking at jumping up to an 8 bus mixer and that’s starts getting much more expensive.

And really when it comes down to it, if you’re setting up a 5.1 workstation, you don’t need or want individual EQs on every channel. And individual volume controls become a big hassel. After spending a long time calibrating the Sound Pressure Level of the room, you want to lock those faders down so that the relative volume from one channel to the next is aways the same. What you really need is what I’ve been refering to as “the box with the big knob” for a year now.

Let’s face it, consumer surround receivers have it right. Plug your 6 channels from a DVD player into the receiver. Six cables from the receiver go to the home theater speakers. And there’s a big knob on the front to adjust the level of all speakers up or down. That’s the idea we need to recreate in the professional sound editor market.

I have yet to find a good solution to this but at the NAB show in Las Vegas, I did see a step in the right direction. Mackie has just released a new mixer that they call surprisingly enough “The Big Knob“. (Mackie, feel free to put a check in the mail made out to me–oh yeah, I never filed a copyright on that name. Damn!) I was so excited when I saw that thing in their booth. No individual channel faders, just a big knob in the middle of the board.

Unfortunately, it’s configured for stereo pairs. It does have 6 ins and outs (3 stereo pairs) but you can only tweak levels on two channels at time. This might not be a problem for Left / Right or Left Surround / Right Surround, but the Center / Sub pair could be a little hairy. It’s possible though that with a little level tweaking on the Pro Tools interface, this mixer could me made to work with a 5.1 studio. I told the guy I talked to at the Mackie booth that I thought it was a great piece of gear, they just needed to get cracking on a true 5.1 version.

Amazingly Colossal

Insanity! The NAB trade show is huge. Think of something big. Now think of something that could beat up that big thing. Now triple the size of that. You’re starting to get the picture of just how big this thing is.

We spent nearly 5 hours today tramping around there talking to people, checking out the latest audio and video hardware, and I am beat. We skipped most of the broadcast video stuff. Spent a lot of time in the Audio / Radio section, and barely touched on the “Multimedia” area. We’re going back tomorrow. That multimedia section is where Apple, Avid, Digidesign, ATTO, Discreet, Adobe, and all the other computer-based hardware and software guys are.

I have spent my share of time at the San Diego Comic-Con and I thought that event was big. This doesn’t even compare. More later. We’re about to go to dinner and Cameron has to see the Lakers.

I’m Leaving On A Jet Plane

In just under 3 hours I will be getting on a plane for Las Vegas. The NAB geekfest awaits me.

Apple is making some big announcement today, if I remember the MacCentral news correctly. So tomorrow I should be able to see what ever cool new products they have at their booth.

I’m also planning on checking out Digidesign, mSoft / Soundminer, Zaxcom Deva, and HHB. Fostex supposedly has some relatively new 2 channel hard disk field recorder that will do 24bit, 96KHz for like $1200. I’ll have to take a look at that.

I’m sure I’ll be bumping into half the engineers and tech guys I work. It’ll be fun to catch up when it’s not some kind of mission critical emegency like normal.

Plus we have reservations tonight at Del Monico’s.

My God, It’s Full Of Stars.

This is cool. Beethoven’s 9th Symphony time stretched to 24 hours with some really sophisticated software. (A standard performance of the 9th Symphony is usually just under 74 minutes. The original CD standard was set so that an entire performance could fit on one disc.)

I’ve been listening to the 5th movement. I feel like I’ve just touched the monolith, and now I need to beat on things with a bone club.

Someone’s Knocking At The Door

It looks like we worked out part of the problem we were having with the Digidesign Core Audio Drivers that came with Pro Tools 6.2.3. If you remember from my earlier post, we are trying to use this on an officially unsupported system: PowerMac G4, OS X 10.3.3, Pro Tools 6.2.3 software, and Pro Tools | 24 Mix Plus hardware. I haven’t really used the OS X core audio drivers before. I didn’t realize that there was a Core Audio Setup program in the Digidesign folder. Once we ran that, and selected the proper interface, everything seemed to work fine. There is still one little bug though: both Pro Tools and the Core Audio Driver think that the 888 | 24 that is hooked up to the Mix Plus cards has 16 channels in and out. (Not the 8 that it actually does.) If haven’t tempted fate to see what would happen if I selected Outputs 9 through 16.

I did notice that Digidesign has slightly updated its compatibility page. It still lists the Mix hardware as being in testing, and they still note that there has not been any problems with a PowerMac G4 in early testing. They have changed to information on the G5 though to say that it does NOT work with Mix hardware and that support is TBA.

We’ve started using Soundminer as our sound effects database program and it’s excellent! We spent many hours yesterday letting the program scan our hard drives full of sound effects, and compile a database. Today Cameron pulled some sound effects for a friend and it took only a few minutes to make some selections and then transfer them to a folder. (This was a huge change from some of our recent experiences with Mtools where we’d spend an hour or two just trying to get the software to behave long enough to get the files on to a hard drive.)

One effect that was needed was the sound of someone getting hit in the face with a bell telephone. That’s not exactly the kind of effect that you tend have sitting around. And even though I’m willing to give up quite a bit of my life to my career, getting smacked with a telephone while a microphone is pointed at me is not one of those sacrifices. So the sound had to be built from individual components. A couple of hard telephone hand set slams, a body hit, a punch, and a slight bell ring off made the perfect “phone introduced to head at high velocity” sound. Cameron didn’t find quite the right bell sound, but he had a lot of great old phone rings. With the Soundminer software, you don’t have to transfer the entire sound effect. You can set in and out points for the piece that you want. He selected the tail-end decay of a phone ring, had just that section transfered, and there was the needed ring-off.

On the on the “when are we going to get a paying gig” news front: we have been given scripts to three different movies that studios are interested in having us do the sound on. All of them start later in the year, and it would really be nice to find something that starts up in May, but I’m not complaining. It’s nice to be wanted. As long as we can convince the “powers that be” that we’re the right ones for the job, we should be busy the second half of this year. Keep your fingers crossed.