
By Roger Nichols
A Bunch Uh Vidiots
The topic for this month is digital audio levels for video. This topic was inspired by a question on the EQ bulletin board and my recent experience with the Steely Dan PBS special.
Background
I will assume that all of you are familiar with analog VU meters used in analog consoles and analog tape machines. VU meters measure the average level of the audio signal. There are analog meters that measure peak levels, but they are only a close approximation of the actual peak levels in the music they are measuring. Analog meters gave us a reference for recording and playback levels. If you recorded to far below the reference level you got extra tape hiss and console noise mixed in with your music. If you recorded too far above the reference level you increased the distortion by overdriving the tape and console. The Zero reference on the meter could be set to accommodate various recording tape and console requirements. Since there could be different Zero references on different tapes, tones were recorded on the tape to tell the next engineer where to set the playback and record levels to match the levels of the music on the tape.
Digital recording came along and a new absolute reference point was established. The reference was called Zero and was set at the absolute maximum level that could be represented by a digital signal. It was also established that all digital bit depths would maintain this reference. A full level 16-bit recording is the same level as a full level 24-bit recording or a full level 8-bit recording. The extra bits give a lower noise floor and a higher resolution to the digital signal.
Lets jump back to analog for a second and talk about headroom. Headroom is the difference in level between the reference point and signal clipping. Most analog consoles specify the headroom as +18 dB, +24 dB or some other amount above Zero VU, depending on the design of the console. If you recorded everything at Zero, the headroom would be where the peaks go. Remember that the VU meters are showing the average levels. If there was no headroom built into the system, everything above Zero would clip and distort. This is exactly what happens in digital recording. If you record at Zero on a digital reference meter, anything over Zero will clip and distort, ALWAYS!
To keep this from happening, a level was chosen as the reference point for recording. The most common reference point is 20 dB. If you record music so that everything reads around 20 on the digital meter, then you have 20 dB of headroom before the signal reaches digital Zero and starts clipping. If you were calibrating a digital recorder to be used with an analog console, a Zero tone on the console would read 20 dB on the meters of the digital tape machine. Some engineers use a reference level of 18 dB, and others use a reference level of 22 dB. The reason is the same. Some engineers who were using 18 fond themselves with levels that crept up during the mix and had to trim the master fader down to keep from clipping the peaks. After a while they decided to just start lower to avoid the master fader trim.
Digital Levels and Copying
Because the digital Zero is the same on all digital machines, there is really no need for a reference tone on the tape to calibrate one digital machine to another. Reference tones are useful for other reasons, but are not mandatory for being able to make a digital transfer. When you copy a tape that had a 18 reference, the copy will also have a 18 reference. In the CD world everyone wants his or her CD to be louder than everyone elses CD. Compressors, limiters and magic boxes were all used to make CDs louder. If you played CDs at home sometimes you had to get up and change the level control on your stereo because of level differences between CDs.
In the Video world, the ultimate reference is the modulation of the carrier for the audio portion of the signal. The FCC has established rules that say that a transmitted signal (AM, FM, TV, CB, Ham, etc.) can not be over-modulated or the station will be fined and if there are too many occurrences the station could loose their license. Audio compressors and limiters are in the circuit to make sure that a maximum limit is not exceeded. TV has contained analog audio since day one of television and established reference levels have become the law of the land. Because peaks above the reference level will over-modulate the signal, there is basically no such thing as dynamics as we know it in most television audio.
So, with a reference point of 20 dB in video production, (some post houses use 18 dB or 23 dB) there is rarely ever any audio material in the area between 20 dB and the digital ceiling. What they should have done is used the Zero digital level as the reference point and they would automatically never over-modulate because there is no such thing as a signal above Zero in the digital domain. With an absolute ceiling of 20 dB, there are three bits that never get used. You basically have 13-bit digital audio, as if it mattered to the video guys. (Ok, I know there are a few who care, but to most of them audio is a thorn in their side.)
Steely Dan PBS Special
In February 2000 Steely Dan videotaped a PBS special that aired in May. The audio mixes were to be used for the PBS show and the DVD that was released June 14. For the DVD we wanted to use all of the bits for the best performance, so the levels were CD levels. The mix was as hot as possible without going over, just like we mix records. We printed the digital audio mixes to the digital audio tracks of the DigiBeta videotape. We gave the same tape to the PBS guys and the DVD guys. We thought (mistakenly) that the audio level would be brought down for the PBS broadcast to match their reference levels. We actually thought that someone would look at the meters on the machine and see that they were peaking at Zero as we stated on the video tape box. I guess not.
When the show aired, it sounded the same as if you ran your favorite CD through a compressor with the compression set to infinite and the threshold set to 20 dB. It was just one giant sucking sound. It would have sounded great if they had turned it down. I thought I was going to have to leave the state on a fast train.
A Quick Flashback
The first digital album I recorded was Donald Fagens "Nightfly" which was completed in 1982. Donald was to do Rock Line, a syndicated radio show and play some of the new album. I had transferred the digital mixes to a Sony PCM-F1 machine and took it to the studio in LA where Rock Line originated. Because of the Zero limit of the digital recording which guaranteed no over-modulation I first sent a full digital level tone over the satellite link to all of the affiliate FM stations that were to carry the broadcast. They all set their levels to that reference tone. During the show Donald was interviewed and cuts from the album were played from the F1. Just before the playing of each piece of music every radio station bypassed all of their processors that protect them from over-modulating. The digital audio was broadcast with no satellite or radio station compression or limiting. Thousands of Rock Line listeners called their stations to say that it was the best sounding music show that they had ever heard. After recent checking, none of the stations have ever turned off their processors since.
The Moral
The moral of this story is never trust anybody to do anything they are supposed to do. You cant count on the guy at PBS to match the levels, you cant count on the FedEx guy to get your master tapes to the record company, the car wash not to scratch your car, the travel agent to get your airline tickets right, the weatherman to tell you if it is going to rain, the accountant to keep you from being audited by the IRS, the drive-thru to give you the correct food order, or your mistress not to call your house and hang up when your wife answers. I guess you just have to move to a desert island where you have to do everything yourself.
Roger Nichols Digital Mastering, No Static Atoll, South Pacific.
Return to EQ index