Miniaturize the Mastering Engineer
I think the hardware end of the playback chain is way ahead of the production end. We’ve got incredible quality DACs from a variety of high-end companies. Just pick your price range and you can get a really good unit for less than $300 dollars like the Meridian Explorer or AudioQuest DragonFly or splurge and get a state-of-the-art unit like my favorite the Benchmark DAC2 for around $2000. The specs on these units are amazing and far exceed the demands that most music productions will ever make on them.
I spoke to John Siau and Rory Rall at Benchmark the other day about the AHB2 Power Amplifier, which has just now being made available (I’m hoping to get one for review soon). This $3000 power amplifier can approach 130 dB according to their website giving it the ability to match the dynamic range of real high-resolution recordings. The power to handle rim shots, orchestral tuttis and instantaneous instrumental or vocal amplitude changes is finally within our reach. Hook a couple of JBL M2 Reference Monitors (if it’s possible…I still have to check on the digital crossovers etc on the Crown amps that are used with the M2s) to the tail end of the new Benchmark AHB2 and the entire playback chain is multiples better than the best sound we’ve ever had. Unfortunately, music production has been moving in the other direction.
If you haven’t downloaded and listened to the examples of the “Mosaic” track by Laurence Juber with the 5 different mastering levels applied, please do so and let me know if you hear the differences. Then think about this…all of the fancy tools for tweaking and compressing dynamics etc. are ALL processes that could be applied at the point of music delivery. They don’t have to be “baked” into the file. Yep, the age of personalizing your music to match your preferences AND your listening environment is upon us…if the labels will go along. And I doubt that they will for reasons that I’ll talk about at another time.
A reader’s comment about mastering and the need to adjust the dynamic range based on the typical listening environment made some very good points. But it was based on the assumption that a mastering engineer has to create a single “one size fits all” version for delivery on a CD or download site. That’s not the case anymore.
Figure 1 – A multiband mastering compressor made by Izotope. This type of signal processor can be automated and built into a file player.
The power of the DSP processors in a music server or SmartPhone could actually perform whatever level of “mastering” you want as part of a realtime process. If you listening at home using HTC Connect in full high-res to connect your HTC M8 HKE Sprint Phone to your high-end system you could choose no compression or light mastering. If you’re on a plane or riding the subway, ramp up the “loudness” wheel on the player app and the dynamics will be tailored to the noisier environment. Who needs a mastering engineer when most of the processing can be stored as metadata and used only when desired or necessary?
Of course, this would mean that the secrets that the mastering folks use while working on the latest hit record would have to be recorded and provided in some sort of standard protocol which the player’s could use to recreate the adjustments in real time. It’s not crazy hard to make the technology work. What’s going to be difficult is changing the minds of the entrenched mastering folks, the artists and the label executives.
They aren’t going to allow it to happen…yet.
DBX used to make some dynamic range expanders. I had a 3BX for years and really liked it with most rock & roll and blues. Nobody seems to make anything like that anymore. Of course, we should not have to expand our music if the recording labels were not over compressing the dynamics in the first place. I like the idea of having a high dynamic range recording and then being able to add compression for certain listening situations. My home theater AVR has the ability to add compression for late night low level listening so you can still hear the soft parts without disturbing anyone sleeping with the loud parts. An automobile is another environment that would benefit. My truck has a speed related gain control. That is not really a compressor but it does boost the volume to overcome increased road noise.
Yes, such thing has been possible since…. 2007ish. Even before the iphone, stuffs like ipod had enough processing power to do basic dsp such as loudness adjustment on-the-fly (with custom firmware such as rockbox)
That said, so called ‘new’ technology by music labels is incredibly old, outdated technologies. Like lossless compression technology FLAC, for instance, has existed since 2000. Version 1.0 came out in 2001, and last major improvement upgrade, 1.2.1, came out in 2007.
It was pretty much matured and completed at that point, so the original developer stopped working on the codec. It was semi-abandoned until Xiph.org took over the development, doing small stuffs here and there.
As I said multiple times, all of these changes should had happened about 5~7 years ago when download was still a thing. Not in today where majority of people are listening to music on Youtube, enjoying incredibly low 96kbps AAC quality (not even bothering Spotify or Pandora at this stage).
Mark,
This is very similar to what I suggested a few weeks ago – simply put a thumbwheel (I know, “old school”) or a digital control that the listener could use to make louder the quieter passages when listening in a noisy environment. This would essentially be a compression control or, to make an analogy to the controls in photo editing programs like Lightroom or Photoshop, a control to brighten the shadows, which would imply that it could be designed to work only on sounds below a certain threshhold volume in the source.
Dolby A noise reduction, anyone? Back in the day, that was too expensive to include in home equipment, leading to the single-band high frequency only “Dolby B,” Dolby A it would be trivially cheap to include now – and all you would be inclluding would be an encoder.- the decoder would be the listener’s mind.
In fact, many home receivers or optical disk players include a ‘nighttime mode” or even – at least on Blu-rays – a control track that can be used to control overall dynamic range.
So give us the unsquashed mixes and let us tailor the compression to our surroundings!
The Crown amplifiers do have analog outputs, but you’d be substituting their 1250 W per channel for only 100 W. from the Benchmark. That would reduce the system’s maximum SPL from 123 dB to 112 dB. You would also get 112 dB by powering a pair of 92 dB sensitive loudspeakers with a 100 W amplifier. Assuming your listening room has a noise floor as low as 30 dB, that gives you an acoustic signal to noise ratio of 82 dB. An amplifier with SNR of 130 dB is nice, but real-world loudspeakers and listening rooms limit its practical value.
Andrea…thanks for the information. I haven’t have a chance to drill down into this yet. But it sure is interesting that we’ve entered an era where the delivery systems are reaching new heights.
Hi Mark,
so, when you listen to playback at loud but pleasurable levels, what peak dBA levels *per speaker* are reaching your listening seat? I bet it isn’t over 105.
I think the quietest my room ever gets to is 40-ish — late at night, and I have a single dwelling at the end of a cul-de-sac road, and inaudible neighbours.
So, how many dB of dynamic range to you suggest I need? Even allowing for, say, 6 dB margin?
When I’m in my studio I listen at 85 dB. But what’s the point of the question? Do you want to accept limited dynamic range because of musical reason, acoustics reasons or at the whim of a mastering engineer? Or maybe it’s better to capture everything that was present in the room where the recording was made and deliver it as it was. Let the circumstances decide how much dynamic range there should be. CD do it quite well…for example.
I’m wondering where do you draw the line, or will you argue to capture at 32 bit, then 48 bit, when the technology arrives? Because ‘it’s better to capture everything’?
The digital systems that I’ve been using for about 25-years internally process at 56-bits. However, as I have stated here before…24-bits is sufficient to meet the dynamic range of human hearing and is therefore enough for audio capture and reproduction. There is no need for 32 or 48 bits in an ADC or DAC.
Very interesting article, Mark! I am an HD Audio fan.
I am intrigued by your idea of including dynamic range adjustment within the player, and providing different dynamic range settings for different purposes, but would stop short of calling it a “mastering” setting as that is spreading misuse of the term. There is no such thing as “light mastering”, for example.
Audio mastering is less about applying dynamic range compression – particularly multi-band compression (shudder), and much more about quality assurance. Obvious dynamic range compression over and above what has already been used by the mix engineer is usually unnecessary as a mastering process, but if the listener wants to add it after the fact, who am I to judge?
For this approach to work to its full potential, I recommend that the compression method and settings be program dependent which is easily doable. By this I mean, a 10ms attack, 100ms as pictured above, is not ideal in all cases. A program-dependent compressor that auto adjusts itself based on the tempo and dynamics of the music would be a better option, which has a better chance of sounding more transparent on a wider variety of program material.
If mult-band absolutely must be used (and I’d argue against the need for that), I recommend keeping all of the compression ratios the same.
Cheers!
Rob..thanks for the comments. However, I’m not sure I’m following your thoughts exactly. Perhaps my definition of mastering is different than yours. The “mastering” stage encompasses a number of processes or tasks. They include ingestion of the source material (possible analog to digital conversion or sample rate conversion), sequencing the tracks, adjusting equalization and dynamics (which usually means doing some degree of compression) and then the entering the PW and and ISRC info/codes prior to a rundown and dump. Having mastering literally thousands of albums for CD and vinyl LP release, there is nothing wrong in my lexicon of saying that something was “lightly” mastered…referring to the amount of compression or dynamic adjustments being performed. There is no reason not to use the term “mastering” even if the processing is done at the final stage of consumption…at least in my opinion.
In my shop, the QA stage was always done by a different engineer than the person that did the EQ and compression etc. That stage is all about making sure that there are no problems or dropouts on the master. I’m been very glad to have someone back me up on occasion.
I’m advocating using metadata developed at the “mastering” studio that is turned on at the delivery device dependent on the needs of the end user and the circumstances of their listening. Multi-band, reverb, ultramaximizer and other processes are among the tools that mastering guys can use or not when “finalizing” a recording for release.
The Isotope company makes very innovative processes/plug ins. My use of the their tool as an illustration was not meant to imply that the setting were anything specific to a tune.
Hi Mark,
I misunderstood your intent. I mistakenly thought you were suggesting that the listener choose between arbitrary “light”, “medium” and “heavy” mastering settings much like the tone controls on a preamp. Having the mastering engineer key in song-specific metadata for optional post-delivery processing sounds intriguing to me, as well.
We will have to agree to disagree on the choice of calling it “Mastering”. I understand that the engineer is still directing what the processor is doing after the listener activates it, but I would still vote to call it “Delivery Processing”, or maybe just calling it what it is – “DSP”, so that there is no risk of creating the impression that the song is compressed because they have “turned the mastering on”. After all, it will only be louder because the engineer has keyed in metadata that makes it so, for a specific listening purpose which may be far more processing than they’d choose on average.
I agree with you about the other processing that could apply. I can envision the engineer using the metadata for equalization, or reverb, while still opting to commit their favorite analog processing to the master prior to delivery because its sound cannot be recreated to their satisfaction by the DSP on delivery.
This idea is definitely worth trying this with AIX Records, in my view. With a library of such pure, exceptional quality recordings, what better place than this to give the listener the option to apply processing to suit their listening situation?
Best of luck!
Rob
Hi again Mark, I wanted to add that I have nothing against Izotope, they make a great suite of products. My comments were intended to illustrate that using multi-band compression with fix-settings – as pictured – is not the best approach for every song, or every genre. Use of that type of tool, using those settings, would very much go against the intent behind having high definition audio (i.e. full dynamic range, wide bandwidth) for a variety of reasons.
Cheers!
Mark,
This is precisely what I advocated a few weeks back in another post. Without a doubt this would make an audiophile quite happy. It might confuse the casual listener but defaults values could make it all transparent to those not interested in changing the sound. I guess the hurdles would be cost of hardware / firmware and getting agreement how it should be implemented. Agreement should be quite painful…