Search this Site

What pros say about it : D. Blanc-Franquard

You're here : Audio >Mastering


worked with Stéphan Eicher, Niagara, Sinclair (his son)...

Before the blessed CD era, we were transferring our delicious 1/4" or 1/2" master tapes on vinyle records. I take this opportunity to mention, with some emotion since this brings us back to Middle Ages, that before bringing the tape to the "mastering" facility, the engineer of that time had to work overnigth to clean his mixes and create, along with the director, the playlist for tracks which were carefully cued together on a large reel with 3 to 4 seconds blank sections in between each title.

When misfortune and the absence of Dolby systems would bring a bit too much noise, hum or polution to the master tape, the producer used to yell when listening to the final cut with headphones, and force us to replace all blank sections with RECORDED tape with equivalent noise to that present in the mixes, so that the listener wouldn't hear absolute silence between tracks... O what a time...

After having been listened to 200 times and finally approved by the prod, the master tape then took the way (of the Cross...) to the mastering house in order to be transferred to the original plate: a metal plate covered with a thin layer of vinyle which a malicious chisel would engrave with more or less success.

Unfortunate ones such as myself who SUFFERED these hard times were marked for the rest of their lives. The engraver was giving us a sample, which we called "flexible" but actually wasn't in any way since it was from the same metal as its twin plate (I still hold a hundred of these in my personnal archives, this nightmare ended in 1991) and was used as a reference when A/B-ing with the first pressed record.

For the sake of memory, let me remind that vinyle records were engraved in stereo using a lateral movement of the chisel for SUM (left + right) information and a vertical movement for DIFFERENCE (left - right) information. However, though lateral movement was only limited by the overall duration of the record (wide grooves: short duration), vertical movement was limited by the thickness (I should say thinness) of the minute layer of vinyle over the metal plate.

SO :

  • impossible to pan the kick drum to the left and the bass to the right
  • impossible to place off-phase information at mid-level
  • impossible to get a DIFFERENCE level higher than the SUM level

Impossible, impossible.... a true nightmare.

Fortunately, relentless and competent people, knowing their machines better than the manufacturer, managed to make miracles, and reference vinyle records that we still listen to with extreme pleasure.

But at ONE condition :

It was essential that recording and mixing had been handled with extreme care and quality for the resulting LP record to be correct.

Otherwise, the talented craftsman would be throwing us out of his lab, nicely but firmly telling us to go learning our job as an engineer... "Sorry old chap I can't engrave this, you've got to remix it..." Absolute shame. Dramatic downfall in front of the director who, with a single burning look at you, made you understand you were fired, burnt to death for the rest of your life, and that you'd better take the first flight to Bengladesh.

I don't exagerate it. A few bunch of colleagues lost their reputation on the road to engraving studios. Others, including me, earned theirs through the engraver's magical words: "I bow to this mix, nothing to touch, it engraves as smoothly as in butter".

These people were loving us because they could enhance our work and do their own properly. They knew that the record company's manager wouldn't be calling them three days later to insult them because the sample pressed record was inaudible when compared to the tape copy they had in their office.

And then some day came CD.

We were told: "Cheer up pals, nightmare is over. No more defects, no more pressure and stress.

The master tape directly goes to the factory, no longer any human intervention involved, the CD goes out and is a perfect clone of the master tape."

The nightmare was actually only just starting. The engraver had discarded his Neumann machines, had come up with Sony mounting benches and the largest swindling of the century of audio finally started...

So, no more engraving, but PRE-MASTERING.

To cut a long story short: you create your mix in studio, on analog tape, DAT, or whatever.

Then you go to the mastering house, where an individual puts the tracks in the right order with complex digital equipment. At first it was the Sony system based on PCM 1630 with Umatic tapes, which is still fairly common. If the original mix is analog, you have to copy it to a Umatic tape, before mounting it on another Umatic tape which will be the final master shipped to the factory.

If you mixed to DAT, you also have to copy it to Umatic where final cut is performed (here we talk about the early days of mastering). Then align levels of each track, possibly correct and limit them. Once cutting is done, you have to encode the Umatic to provide the PQ codes.

PQ codes are used for indexing the CD. Thanks to them the CD player finds where the beginning of each track is. There's an index at the beginning of the title, and one at the end even when titles are chained without pause between them. The end index is used for computing the duration of the track. The CD player is able to browse the CD thanks to them. Without going into boring technical stuff (especially boring for us when we have to do it), beginning and end indexes have offsets of a few milliseconds, allowing for the player to position slighly ahead of the beginning of a track so as not to eat out its intro when it starts playing. The Red Book standard is very strict in this respect. Any master tape going to the factory which does not scrupulously abide by these rules is mercilessly rejected. In order to make sure these rules are properly followed, the Umatic tape and recorder go through an ANALYSIS which is printed out on paper and provides all the necessary details: number of errors corrected, comparison between written and descibed PQs, exact number of written data blocks, etc. They don't have fun at all in these factories.

The master tape reaches the factory. It's put into another Umatic, and another analysis is performed in order to make sure that the machine playing the tape does not generate more errors than the one which recorded it. If everything is OK, the Umatic is inserted in the LBR (Laser Beam Recorder) rack and the Glass Master is engraved. Then, it's over, robots take over the remaining process and round-shaped end products pile up in crystal plastic cases, a press producing nothing less but 500 CDs at once. No single unit samples, your CD directly ships to the record company's warehouse and from there to the retail channels. The smallest error goes affecting all pressed CDs, without the slightest chance of control. All this story was taking place before virtual became the name of the game.

Nowadays, mastering on Umatic is almost as outdated as steam locomotives, for the mere reason that since virtual came into play, it was discovered that you could engrave the Glass Master at double speed when using computers. Double speed = twice as many Glass Masters, twice as many presses and thousands more CDs produced per day...

Today, depending on the factory, you can get a Glass Master from such sources as Umatic, DDP tape (8mm Exabyte data tape including a disk image of the CD (the current standard), PMCD (master CD including same information as a Umatic), or a mere CD burned at home.

Pursuant to information described above, it therefore appears that in order to get to the DDP player which is now the standard equipment in the LBR rack, any other format must first be converted to DDP.


That there still is some human intervention involved after the pre-mastering phase and that this intervention, when performed without proper competencies, can simply ruin several months of work. But we're getting off-topic, let's keep this for another debate.

So then, is it necessary to go through the pre-mastering house to ship one's tape to the factory ?

As eveyone knows today, CD is using 16-bit format. Due to the various required signal processing involved if one has to compress, equalize or boost the master, audio files used for pre-mastering are USUALLY 24-bit ones. I say usually, besause in absolute theory, copying to 24 bits a crap analog tape which mimics the speaker's membrane of a 100w Marshall just coming back from Metallica's world tour won't probably change the face of the world, but anyway...

24 bits is the standard in pre-mastering houses, where audio files are processed with SADIE on PC or SonicSolutions on Mac computers. Umatic is 16-bit, thus its Beingg outdated now. Once cutting is completed, with eq and various processing performed, the final master is re-built by going down from 24 bits to 16 bits using complex dithering algorithms such as Apogee's UV22, or Sony's Superbit Mapping. An audio CD can only contain 16bit/44.1kHz data.

When the CD first appeared, in 1982, the standard was stating that the CD's standard max level, i.e. 0dbfs, was 18db above zero vu on the mixing board. In order to keep some headroom for peaks. As years went by, we've now reached in 1998 the point to which on current CDs this headroom is reduced to 2, even only 1db. In other words, on these CDs the average level is 16db higher than it was on in 1982 CDs.

The level war is endless ( :-) ) and each mastering studio competes, as during 45 rpm singles times, in order to produce a master even louder than the one from the studio next door.

Even if it's bearing some effect, there's however an issue, because the average CD player, having been conceived for a 18db headroom, finds it quite hard to deliver 0db average level from its cheap preamp section. It even happens that the CD's level is such as to impose to the player to draw more power from its main than the latter can deliver, resulting is distorted, horrible sound.

The fact of having high level on a CD is not linked to the fact that it was mastered using 24-bit resolution, but because the audio material was limited and compressed in such a way as to remove all peaks, therefore allowing for rasing the level by 3, 6 or even 9db without trouble. Needless to say that if the music wasn't conceived accordingly, the final result may be somewhat disappointing from an artistic standpoint. Louder doesn't always mean better.

Anyone can now make his or her own master at home, with a little care. Equipment such as the TC Finalizer, when properly used, can produce the same theoretical results as those we get for a pro mastering. One mostly needs some experience about these techniques, because any error is of course fatal. Because it's difficult, without specialized gear, to be sure to really abide by the factories' standards.

As far as the mix is concerned, you have to be sure of the desired variations in level. When the CD is made to be listened to by a normal person, without having 2 x 1254w hi-fi equipment installed in a concrete bunker totally isolated from surrounding noise polutions, it's quite hard to enjoy variations higher than 6db without having to endlessly raise the volume to hear the ppp, then lower it when coming across fff passages, etc.

If you're 100% sure and were able to verify that contemplated variations are REALISTIC, just try to nicely limit the sharpest instruments, those which make the peak meter reach 0, even if you feel there's not enough loudness. In this list, ALL drums, some piano notes, vocal consons (Ts, Ks... too much compressed).

If each instrument was carefully "trimmed down" without altering its quality, the mix will at least show a peak/level ratio closer to reality. If for any reason an instrument spews a peak 6db louder than all others, the average level of the CD will be 6db quiter if this peak is the 0db reference. Too bad...

Once the mix is completed, it's a matter of carefully checking it up; to this end, mastering consists in converting it to standards of the outer world. Meaning: equalizing the master, if necessary, in order to level off defects of the monitoring system used when mixing it. Obviously this can only be performed using reference, long tested monitors on which everything is audible, imperfections as much as qualities of a mix, i.e. which were already used when casting out some major hits.

Then the overall level of the master needs to be boosted, if possible, so that when the CD is released, it compares to other productions around the world so that the only differences between it and them are musical, not technical ones.

Boosting means increasing level, peaks allowing. If the master was produced on a non normalized DAT, there's some headroom left. Boosting also means raising the noise floor which hides behind the musical material. If the mix is "inconsistent" level-wise, it will have to be compressed. Levelling off inaudible peaks does not mean overcompressing the mix, just raising overall level. Analog tape, for that matter, used to have (and still displays) the talent of softly doing this job.

The human ear hates peaks, which are naturally levelled off by the eardrum. Leaving them brings nothing from a musical standpoint, but a blatant sensation of aural weariness after some minutes of listening. A "sharp" sound, deprived of thickness. Of course this processing degrades the mix's signal to noise ratio. That's why the mastering standard was set to 24 bits, in order to leave more room for these processes.

But as long as the musical content is enhanced, the end result is Beingg optimized. Who minds hiss and distortion if the music sounds better? The only thing that matters is the end result. When doing this job alone and at home, the only piece of advice I can provide is to compare one's own mixes with tons of well-kown CDs which approximate the musical content on which one is working. In order not to be deceived by cheap D/A converters, try putting a hand on a CD player with digital outs. For Pro Tools, the less expensive solution is to use Waves' Maximizer plug-in. With some tweaking, you can easily find quite convincing level optimization settings.

Top of page
Texte publié avec l'aimable autorisation de Dominique Blanc-Francard - DBFMusic/Labomatic Studios, on the 21-05-1999

Page viewed 11262 times