Jump to content

Technical Question About "Sunset Strip"


FloridaPilot

Recommended Posts

I just noticed something interesting... on the DVD there is a big difference in sound quality between the PCM stereo tracks and the AC3 surround track.

Up until now, I had listened only to the surround track and I thought it could have been better.

But I just started watching it again from the stereo track and there is much more presence to the entire mix. I can a-b the two tracks at the touch of a button with my DVD player.

What I am hearing is more than the inherent difference between uncompressed PCM and highly compressed Dolby Digital.

I recommend that anyone who has only listened to the surround version to try it in stereo.

I am listening through a Yamaha DTS-ES/Dolby Digital-EX preamp/processor, a rack of ADCOM GFA-5500 amps running Martin Logan Aerius i front and rear and the ML Center, and an SAE 7600 running a dbx dvc sub.

Link to comment
Share on other sites

I was thinking perhaps there might be other audiophiles on the board.

In case anyone is interested, here is a bit of info on digital audio & compression, PCM versus dolby digital, what is wrong with CD and what is horribly wrong with mp3 and Dolby Digital surround. PCM is uncompressed digital sound, like what is on a CD. The basic resolution of digital sound is determined by 2 things- Quantization and Sampling Rate.

This can be explained using a movie film analogy. Movies in the theater are typically 24 frames per second, on 35mm film prints. The "frames per second" is analogous to sampling rate, and the resolution of each individual 35mm frame is analogous to quantization. With film, increasing the frame rate increases the smoothness, or the illusion of fluid motion, whereas increasing the film to 70mm (like IMAX) would increase the resolution, or detail, in each frame. In digital audio, Quantization is the resolution of each sample, or frame, of audio. Sampling rate, or the number of frames (samples) per second, determines the highest pitch that can be recorded.

In audio, pitch is called frequency, or how frequently (in cycles per second) the air vibrates. The human ear can hear from about 20 vibrations per second to about 20,000 vibrations (or cycles) per second. Cycles per second are called "Hertz" or Hz. So, the range of human hearing is about 20Hz to 20kHz.

Here's how sampling rate determines the highest pitch that can be recorded digitally. The Nyquist Theorem states that the sampling rate must be at least twice the frequency as the highest pitch you want to record. Quite simply, this is because it takes a minimun of 2 points to determine one cycle of sound- the top and the bottom point. If you drew one cycle of a sine wave, you could represent it with two points- the top and the bottom. You could connect the dots to draw the wave (although it would come out a triangle wave- more on that later.) So, if you wanted to create a digital audio format capable of reproducing the entire pitch range of human hearing, you would need a sampling rate of at least 40,000Hz- the 20,000Hz limit of human hearing times two.

Therefore, CD sampling rate is 44,100 Hz.

The extra 4100 Hz is necessary for filtering to prevent aliasing. Aliasing in film is the "wagonwheel effect" where a wagon is moving forward but the wheel appears to be moving backward. This is an artifact of where the wheel was when that particular frame of film was captured. If the film frame rate were increased beyond 24 frames per second, the wagonwheel effect could be eliminated up to a certain wagon speed.

In digital audio, aliasing creates phantom tones at lower pitches, called "foldover" frequencies, whenever any frequency above the sampling rate is sampled. So, in reality, sampling rate must be higher than the Nyquist frequency to allow room to filter out any frequencies above the sampling rate. That is why CD is 44,100 instead of 40,000 sampling rate. CD Quantization is 16 bits. This is analogous to the resolution of a still image, or one frame of film.

So, CD audio consists of one 16-bit sample (frame of audio) 44,100 times per second.

Therefore, CD "bitrate" is 1,411,200 bits per second, or 1400 kbps (thousand bits per second). This is determined by multiplying 16 bits times 44100 samples per second, and multiplying the total by 2 (left, right). The bitrate of digital audio indicates the resolution of the audio. Higher bitrate = better sound and larger file size.

THE BIGGEST PROBLEMS WITH CD

  1. 16 bits does not provide adequate resolution. There is a notable difference between 16 bit sound and 20 or 24 bit sound. Each added bit doubles the resolution of each sample. So, 20-bit digital audio has 16 times the resolution per sample as 16-bit audio. That would be like taking the 35mm film print and expanding it to 16 times the area- way bigger than even an Imax print. CD was designed only to be a mid-fi product, but was marketed as the greatest thing since the advent of recorded sound. 16 bits provide 65,536 different values. when the actual original sound falls between two of these values (which happens all the time in digital audio) ,it gets assigned the closest one. This is called quantization distortion and is very audible at 16 bits. This is similar to "pixellation" effects in low-resolution digital images.
  2. Sampling rate is way too low. In reality, sound well above the range of human hearing should be recorded for true high-fidelity. Why record sound that the human ear can't hear anyway? The answer is that frequencies above our range of hearing affect the tonal characteristics of the frequencies we CAN hear. Harmonics (higher-frequency overtones) are what makes a G note played on a piano sound different from a G played on guitar. Harmonics give a tone its distinctive sound. If you have ever played with an old analog synthesizer, you can set the wave shape of a tone (square, sine, triangle, sawtooth, etc.) for a given note, each of these wave shapes sound notably different. This is exactly the effect that ultrasonic harmonics have on sounds at the upper limits of human hearing. The inaudible harmonics modulate the audible wave and change the tonal characteristics of the upper end of the audible spectrum. Also, the anti-aliasing filtering mentioned above takes place too close to the audible spectrum in CD-audio. A by-product of this filtering is the "rounding" off of the higher frequencies of the audible spectrum to sine waves. Because the sampling rate is bare-minimum, only triangle waves can be represented at 20Khz anyway.

For these reasons, on CD the uppermost sounds of the audible spectrum are shaped like sine waves, regardless of what waveform the actual original sound had. These reasons are why many audiophiles say CD's sound, metallic, harsh, mechanical, etc. The entire upper range is distorted timbrally. These problems with CD really are quite audible to most people given listening experience. If you have ever been in a modern recording studio, and you are listening to the two-track analog master reel tape of your band's performance, and then you burn it to CD, there is a notable difference in the sound no matter what CD Recorder you use.

Newer audio formats, such as SACD (super audio CD) or ProTools used in many recording studios sample at 192khz and have 24-bit audio. This sounds much closer to analog reel tape than CD.

There are other problems with CD, but these are the two biggest.

Another problem is the recording techniques used to get the best sound out of CD. CD sound too in-your-face and this is the result of dynamic range compression in the studio. This has no relation to digital audio compression. Dynamic Range compression reduces the difference between the loudest portion of a signal and the softest.

As stated above, the 65536 resolution points provided by sixteen bits ain't enough. Compounding the problem is that the lower the volume level of a signal (either the entire mix, or an individual instrument in a digitally tracked recording), the less of those 65,535 bits are available. The full 65536 are available only at the highest possible recording level. So, the lower the volume level of an instrument or overall mix, the higher the quantization distortion.

On the flipside, with digital audio, the 0db recording level (the highest possible recording level and where quant distortion is the lowest) is a brick wall. If the signal exceeds 0db by one iota, you hit a brickwall of distortion because you have run out of bits with which to represent the signal. With analog tape, exceeding 0db caused tape saturation, which is a very pleasing effect to a point.

So, digital recording engineers or mastering engineers typically compress the shit out of everything so that the lowest volume levels are still pretty high where quantization is reduced, but that the signal never exceeds 0db. This causes CD's to be too in-your-face, unnatural and fatiguing to the ear after a while.

COMPRESSED AUDIO

CD-bitrate files are very large. One way to reduce the file size would be to lower quantization, but when you go down to 8 bits things start sounding pretty horrible. You could also reduce the sampling rate, but then the highest frequency you can record goes down and your treble range disappears. By the time you substantially reduce file size, you wind up with something that sounds like a telephone or AM radio.

Digital Audio Compression was designed to reduce the file size while still being able to record higher pitch sounds and still having reasonable (!?) resolution per sample. Mathematically lossless compression is identical to the original with no loss. However, lossless compression only about halves the bitrate. So, if you are listening to digital audio at a bitrate below about 700 kbps, there is a difference. Lossy compression formats such as Dolby Digital, MPEG Audio, mp3, etc. are way below this level, do NOT sound as good as the original, and are NOT CD quality regardless of what the manufacturers tell you.

Generally speaking, with digital audio compression, sampling rate and quantization are not altered. Instead, psychoacoustic principles are applied to permanently throw away bits of the audio. This is why i say it should be called "data REDUCTION" instead of data compression.

There are a few ideosynchrasies of the human ear that are exploited.

  1. threshold of human hearing. Sounds below a certain volume level are not perceived an can be thrown away. The threshold of human hearing also varies with pitch and volume. At lower volume levels, we percieve the bass and treble less. That is why old stereos used to have a "loudness" button, which was short for "loudness contour compensation". A true loudness circuit would boost bass and treble at lower volumes, and gradually decrease the boost as the volume was raised. This allowed the human ear to hear a balanced signal at low volume. Lossy compression algorithms use this threshold of hearing curve to throw away audio depending on their relative volume and frequency.
  2. HAAS effect. In a nutshell, if you hear the same sound twice with just a few milliseconds between them, the brain does not percieve the second sound. This is why older Dolby "Pro-Logic" surround receivers had a digital delay feature. The purpose was to eliminate the perception of center channel dialogue leakage into the rear speakers. It was there, but if you set the delay right you couldn't hear it. Compression algorithms will look for patterns in the audio that resemble this (are within the HAAS range) and omit sounds accordingly.
  3. Masking. In human hearing, louder sounds mask softer sounds. If you are standing on the sidewalk of a busy intersection with horns honking, engines running, etc., and someone is talking across the street, you can't hear it even though those sound vibrations are hitting your eardrum. Compression algorithms delete quieter sounds at the same or similar frequency as a louder sound.
  4. Dynamic bit allocation. For example, if there is alot of activity going on in the midrange of a song (a screaming Wally Bryson guitar solo, for example) more available bits would be allocated to the midrange, and less to the treble. It is harder to perceive detail in bass frequencies, so the bass is usually always given less bits than the rest of the spectrum, and the amount varies from one compression format to another. Some formats (like early versions of Sony's ATRAC compression) gave very few bits to the bass, and the bottom end lacks impact.
  5. Brickwall filtering. Often times, especially at lower bitrates, the upper octaves of the audible spectrum are deleted altogether. Frequencies high enough to give some degree of crispness are retained, but everything above is gone resulting in a constrained, coarse sound with a complete lack of space, soundstage and airiness.

It is theoretically possible to use these techniques (except the fifth) to imperceptibly reduce file size, the problems are twofold. One, creating the perfect algorithm is like finding the exact value of pi.

Two, even if that could be done, the maximum about of reduction possible would be a little over half, or down to around 400 or 500 kbps.

MP3s, etc are WAY lower than this. The average MP3 file is encoded at 128 kbps, which is often represented as "CD Quality" when in fact over 90% of the audio from the CD has been discarded.. The other thing to know is that some compression algorithms are much better than others. MP3 is among the worst. At a given bitrate, WMA sounds about twice as good.

So, when you listen to compressed audio like mp3, you are not getting close to CD quality, and even CD is a mediocre format. Satellite Radio (both XM and Sirius) compress down to around 50kbps- so frikkin bad it is like fingers down the blackboard and I can't believe anybody can stand it.

DOLBY DIGITAL

Also known as AC-3 for Audio Compression type 3. This is the gadawful format chosen as standard for DVD. It is a low-bitrate lossy compression format. It allocates around 320 (don't recall exact #, but it is in the 300's) kbps total for ALL FIVE Channels plus the subwoofer channel! So less that 20% of the bit rate of CD has to cover 5.1 channels!

Not only do each of the channels in and of themselves sound dull because of this, but if you actually read Dolby Labs whitepapers on the format, part of the compression involves blending high frequencies together in the surround channels! So, what is supposed to be 5.1 discrete (separate) channels really isn't. Blending the high frequencies of separate channels gives Dolby Digital a major lack of space- it sounds constrained and lifeless.

For this reason, people that care about the listener will give a PCM (uncompressed CD-quality stereo) track on music DVDs, as Mr. Linett has done for us.

On "Live on Sunset Strip" choosing "Stereo" in the audio setup menu selects the PCM audio track.

DTS. You may have seen DTS on some DVD's. This is a competing format that was actually the first multichannel digital sound format for theaters. First film was Jurassic Park. DTS is a much higher bitrate than Dolby Digital- about 1200 kbps. This is still lossy compression, but sounds MUCH better than DD. Unfortunately the DVD consortium chose the crappier format as the standard.

I have a beef with the misinformation peddled by electronics manufacturers today.

First they told us CD was the best possible sound, when even 25 years ago that wasn't true. Then they told us that MP3 at 128 kbps is CD Quality, which is a laugher. They told Dolby Digital is 5.1 discrete channels, which isn't quite true. Now they're peddling HD Radio, which is an absolute LIE.

HD radio is very-low-bitrate compressed digital audio. There is nothing "High Definition" about it. In fact, if definition and detail are roughly synonymous in this context, HD Radio is much lower definition than analog FM.

If you were still able to find a broadcast FM station that wasn't playing its music from compressed digital sources, with a quality tuner and a well-engineered station you would find the sound to be extremely good. The noise floor may be lower with HD radio because it is digital, but it is very LOW definition. If they called it LN Radio (low noise radio) I still wouldn't want anything to do with it but at least they'd be being honest.

I'm tired of these companies peddling crap at us and bullsh--ing us into thinking it's great.

There's my .02

Mark- I'm curious, are you an analog advocate or a digital believer? My opinion is I think 24-bit, 192 KHz digital sounds pretty good through a good D/A, but to me nothing touches the sound of 2" analog tape.

Link to comment
Share on other sites

Woah , WAY too much techno-babble for me, and I'm in this business! I always felt when cd's first came out that they sounded like crap, but after a year or two, I was hooked. I myself hear everything I need to hear. I'm thinking that James Taylor's "October Road" or his "Hourglass" albums are good examples of well engineered cd's.

Link to comment
Share on other sites

Hi Jeff,

Perhaps we'll have another chance if the berries tour again in support of LOSS.

Thanks again for the after-party in Chicago, by the way! Hope all is well with you.

MAM- I agree with you about Hourglass- one of JT's best, and definitely the best in 20 years. I think much of that record was recorded on ADATs in a small project studio. I think there are a few albums that make the most of the 16-bit digital format, and Hourglass is definitely one of them.

Link to comment
Share on other sites

Jason wrote

"There's my .02

Mark- I'm curious, are you an analog advocate or a digital believer? "

A bit more than two cents I think........ quite a rundown of the various digital formats. Thanks.

To answer your question I prefer digital as a storage medium and often record at 88.2/ 24 bit, but I still mix through an analog API console which is how the live CD/ DVD was done.

I like to think of Pro-tools as a big ultra-fancy tape deck that allows me to do all sorts of creative things like editing that were laborious or impossible on actual tape.

The quality of analog to digital converters (and vice versa) has been improving steadily in recent years and imo unless you want the "good" distortion that overdriving analog tape can provide, digital can and does sound just as good and in some ways better than analog.

That said I have made records in the "old days" (1990's) on 44.1/ 16 bit digital tape decks that still sound good to me.

This is of course a very subjective discussion, but overall let me say that digital audio, like multi-track tape, and analog tape itself all have made enormous changes in the way music is recorded. None are in and of themselves "better" . They are merely tools that can be used creatively and as I like to say for good or evil.

Mark

Link to comment
Share on other sites

BluRay and HD-DVD use higher standards for sound but very few receivers can decode LOSSLESS and TRUE DOLBY off the High Def DVD formats.

There was a lot of debate on DTS and 5.1 on movies and for most its really hard to tell much of a difference when the DVD's are coded with each and you can switch between them on the fly.

I was a little disappointed that the surround chanels on Live On Sunset Strip were not more active but that is a personal taste for sure.

Some people like to hear only audience in the surrounds but others like the mix to be like one is sitting in the center of the stage and you can hear different members of the band in each of the 5 channels. ELO's ZOOM and The Eagles Live in Melbourne DVD's are mixed that way and they are a mind blowing listen/view IMO.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...