|
Educational Articles
On the Level about LEVEL
or "What is Zero Anyway"?
By Manfred N. Klemme
Occasional Paper Number two for STELLADAT users.
One of the many things that audio folk cannot agree on is level. Some record hot, some down in the mud. Some even fiddle with the "nano Webers" so people using the equipment don't record as hot as they think. Everyone is trying to fit the dynamic range of the source onto the straight line portion of the curve on the tape without going over the top.
When material is going to be used by others, such as when it will be transferred to another medium some reference is needed. The reference should be standard but alas it never is. 1/4" analog recorders used for motion picture dialog recording finally settled on a level of -8dB on the Nagra modulometer as the reference level. This corresponds to 0 dB on a VU meter... sort of. At least it is a standard. The transfer persons simply set the recording level of the -8 reference tone to 0 on the mag film machine and hit record. If the tracks were recorded correctly everything was OK. Whatever dynamics could be fit onto the 1/4" probably fit on the mag film as well.
In comes DIGITAL! There is no way to adjust the recording level on a DAT machine. That is factory set. We are only recording ones and zeros anyway so we may as well engrave them on the tape. What then is the "LEVEL" control on the front on the machine? It doesn't control recording level, it controls how many "bits" we are using. More bits means louder and less softer, but is not related to the level of magnetic flux laid on the tape. The "LEVEL" control is essentially a "Dynamic Range" control. We don't worry about the straight line portion of the curve, There is no curve.
The "art" of recording on a DAT machine is fitting the dynamic range of the source into the available dynamic range of the medium to which it will be transferred. This is similar to the "art" of photography, where the photographer attempts to fit the contrast range of the scene into that which can be reproduced by light reflecting off white paper. Blocked up shadows and burned out highlights show when he missed. We mess up when the delicate little sounds that were so beautiful on the DAT end up in the mud, and the thundering highs distort on the mag film.
So, what is "ZERO"? The SMPTE (or was it the AES) has decreed that -20 dB shall be the "ZERO" on digital tape. Why? They say that if one records so that peaks are at or about the ZERO (-20 dB) there is 20 dB of safety in case someone screams too loud. We must presume the real zero is where all the bits are ones or zeros (I don't know which). At that point there is hellish distortion if there is no limiter. The Stelladat as a fine limiter at about -3 dB so that can't happen. Thank goodness.
In any case, to conform to the standards it should be the practice to set a reference tone at the head of the tape recorded so that the meter on the Stelladat reads -20 dB and then control the level so that the program material peaks at this point. (A little hotter as on the Nagra won't hurt)
One thing to remember is that when the DAT is played back as an analog signal the level can be controlled before it is recorded to another medium. However, when transferring the digital data to another digital device, there is NO control of level. Only the ones and zeros that were recorded are rerecorded. Its like copying a disc on a computer. Conforming to a standard makes sense.
On the Stelladat -20 dB is indicated when the meter is set on "PEAK METER" and the dB scale is set to "DSP nominal 0". Some day we may provide software that has a selection for - 20 dB indicating 0 on the meter.
Manfred N Klemme
SONOSAX USA, Inc. 5417 Cahuenga Blvd. #C North Hollywood, CA 91601
The information presented in this paper is intended to promote thought and stimulate discussion in sound recording techniques. It is NOT to be regarded as standard procedures, or recommended practices. The user of the equipment MUST determine what techniques are compatible with the requirements of the production and post production processes.
c Manfred N Klemme 1994
|
