Daniel Weiss Interview:
"People tend to think that digital is simple, just use that algorithm. But the degrees of freedom in digital algorithm design are also very many, like in analog..."
With Daniel Weiss about the sound of digital, technology, the history of Weiss Engineering, brand new LIVEBOX solution, and the extraordinary software native version of his most famous hardware mastering processor - the one and only WEISS DS1-MK3, which was created by the well-known and respected SOFTUBE Company with the close involvement of Daniel himself...
Daniel Weiss is a man who certainly doesn't need to be introduced to anyone who has ever done any form of mastering, whether analog, digital or hybrid. Nor to anyone who at least once looked for a way to get the best final sound of their project. Although he would probably never say it about himself, the fact is, he is a legend.
And with this bold, though extremely true, statement, this introduction could be completed.
In my opinion, however, there is something else that needs to be added. Something that makes him not only a legend, but such an extraordinary man... His humility, modesty and diligence.
While his legendary mastering dynamics processor revolutionized the sound of music all over the world, he remained firmly on the ground, just doing his job as well as possible. That's how he is today. Modest, objective and reliable, although there is no doubt, that his devices are some of the best audio production tools ever built in the world, although there is no doubt, that it was his DS1 that introduced the digital to the mainstream, and it was his DS1 that seriously and emphatically proved that digital can have its unique and very decent sound, he constantly approaches his professional path sensibly and in a very balanced way...
People with his knowledge, with such a unique combination of knowledge, passion and talent have never been enough in the world, year after year there are also, unfortunately, less and less of them.
That's why I'm even more honored that we could talk about some of his, both well renowned and more recent gear, icluding an extremely promising brand new LIVEBOX solution, as well as the digital sound, the history of Weiss Engineering, and the extraordinary software reproduction of his most famous mastering dynamics processor, the one and only WEISS DS1-MK3, which was created by the well-known and respected Softube company with the close involvement of Daniel himself...
Adrian Lucas Witaszczyk: Hello Daniel, thank you so much for finding the time to talk!
Daniel Weiss: My Pleasure Adrian Thanks for inviting me here!
How did your journey to the top begin, I mean how did your equipment get such a huge reputation? How do you remember that? Who first discovered your equipment? Was it the effect of a so-called snowball or was everything happening slowly?
I started my company in 1984. That was when the CD became popular. This means there was almost all of a sudden a market niche for audio equipment for digital audio signals. I had contacts to a mastering studio in Germany (Harmonia Mundi Acustica, HMA) and as a joint venture we started designing digital audio equipment for Mastering Studios. HMA did the specifications and sales and we did the designing and manufacturing.
In the beginning most of the equipment was sold in the USA as Mastering was invented in the USA and thus very popular already. We had almost no competition back then. Our system (bw102) was modular and thus very flexibly configurable.
We quickly gained a very good reputation.
At the beginning of your own business you were responsible for everything including: designing components, PCBs, DSP software, assembling devices, testing... how do you recall that time? Was it hard? How did you motivate yourself?
Yes, I did a lot on my own back then. Motivation was not a problem. Before starting the company I worked at Studer as design engineer in the PCM lab, so that part I knew. It was, and still is very interesting to design new stuff.
Looking at your career path, you might get the impression that you've always against the tide. When you started almost everyone was blindly holding onto the analog, you went towards the digital domain...
Because I started at Studer in 1979 in the PCM lab, right when Studer started to do digital audio, it was kind of logical – with the advent of the CD – that digital audio is the future. I'll take the risk...
What you were doing at Studer?
I mainly worked at the design of the SFC16 Sampling Frequency Converter. And also at the first DASH recorder. (DASH = Digital Audio Stationary Head). And I also did analog anti-aliasing / reconstruction filters for converters.
Why digital domain? Did you ever try to create an full analog device?
Recordings were made in the digital domain already and thus it made sense to do digital processing right up to the CD master in order to avoid A/D and D/A conversions. A/D and D/A converters were not that good back then.
We have one device with analog in and out – the Series 500 A1 preamp / de-esser. It is almost fully analog, only the sidechain for the de-esser is made with a DSP and a multiplying DAC.
The breakthrough in your career seems to be the BW102 modular mastering system, which you mentioned.
Where did the idea for this system come from? How did you perceive mastering at that time and what exactly were you lacking in the solutions available at that time?
Yes, as mentioned earlier, the requirements for that system were defined with the help of HMA. I had the idea for the modularity. This was essential I think, as the system grew slowly over time by adding modules for various functions.
In the beginning various interfaces were required as there wasn’t any standard like AES/EBU. The manufacturers had their own interfacing formats. Also sampling frequencies were hardly standardized, so a sampling frequency converter was necessary as well. Plus digital level control, a highpass filter to get rid of DC offset and a De-emphasis filter to get rid of Emphasis. Later we added equalizing, dynamics, mixing and more.
If you don't mind, let us now move on to the main star of our meeting today… The WEISS DS1 MK3 Mastering Processor and its plugin version released by Softube...
Yeah, sure, go ahead...
This plugin... This device is so versatile and rich in options and settings that it is impossible to discuss it in one, two or even three conversations! So many questions come to mind... Do you remember the day when you came up with the DS1 concept?
The DS1 grew out of the bw102 dynamics and bw102 de-esser modules. We added some features like soft-knee or that the filters for de-essing span frequencies down to the bass range. This sounds easy, but is not that simple because linear phase filters at low frequencies tend to take a lot of DSP power.
How did you deal with that?
At lower corner frequencies we change the sampling frequency to a lower value.
What directly led to its creation?
We started to do non-modular units (the Gambit Series) and already had the EQ1 equalizer in that series. The DS1 was a logical step.
I would describe the DS1 as a digital processor that has a soul. It sounds a bit like a cliche, but when you spend some time with either the hardware or the software version from Softube, you start to understand what I mean…
This compressor does not seem to add any color or distortion. It doesn't affect the sound of the source material and still has a ton of soul! It adds groove, feeling and excitement…
Thank you for that opinion, well I am glad that this device still surprises and is able to evoke such emotions in new generations of mastering engineers.
How did you develop such a good algorithms and how long have you been working on them? Or you know what...
I will ask you directly... how did you do it, that a digital device from the beginning of the 1990s still beats analog processors to the head?
When we design an algorithm we always try to get the most out of it in a technical sense. I.e. we try to have very good SNR, no distortion, as distortion in a digital system can get very nasty due to aliasing, good frequency response etc.
And we also try to avoid cookbook algorithms. People tend to think that digital is simple, just use that algorithm. But the degrees of freedom in digital algorithm design are also very many, like in analog. Digital is a model of the analog world and even can get beyond that because certain things (delays) are hard to do in analog at audio frequencies.
You mentioned aliasing, many plugins that try to emulate old analog hardware suffer from it. Maybe it's too broad a kind of question, but how is it actually when you create a plugin? How do you eliminate it?
One needs to avoid the harmonics, which are generated due to non-linearities, to go above half the sampling frequency.
So no aliasing is generated. As an example, if I have a 2nd harmonic generator I will need a low pass filter in front of that generator which avoids frequencies above ¼ th of the sampling frequency to reach the generator.
Did you have a feeling then, that it would be such a good device? Did you feel that you were doing something so big? That you making a history?
We knew because of the bw102 modules, which were very well received, that we won’t do much wrong with the DS1.
The attack control of both hardware and software DS1 MK3 has a mysterious “Preview” setting. Can you tell me what it is?
Let’s get back to the basic compressor diagram: A circuit (the so called sidechain) is measuring the envelope of the music signal and that envelope is taken to control the volume of the output signal. Very simple. The problem is that the envelope usually is behind the signal, timewise. E.g. if I would like to compress a fast peak, the envelope detection lags behind the audio signal (attack time) and thus misses to tame the peak. This can be avoided by inserting a delay ahead of the output volume control (where the sidechain is taking action).
Another major feature is the inclusion of reverse compression ratios, also known as expansion.
The expander used in conjunction with the sidechain filter can do really great things, like making dull mixes jump to life or recovering peaks lost at earlier stages of production...
Yes, right. The combination of an expander with split-band processing may be something “unusual”.
When we add to all of this, an amazing attack whose fastest setting is equal to the sample level (20 microseconds) and extends up to 800 milliseconds, three release settings, including a great release delay feature, which allows you to work in a mode reminiscent of a vintage Fairchild Limiter, (in which we have the so-called initial quick release, medium stage release and something that can be described as long term release, which are designed to protect each other so that the gain recovery does not come back too much between the quiet sections while broadcasting) as well, as the ultra clean and probably the most non-destructive de-esser which we can imagine… we will get a device that went very far beyond its era.
It's fascinating, not only - you have to spend a lot of time on it, but you also had to have an incredible knowledge and intuition as to how the audio industry will develop in the next 10 - 15 years... It looks like you've thought of everything…
Well, compression, expansion, de-essing etc. are very old concepts. With a digital implementation there are some advantages available though, like that preview delay as mentioned before or linear phase band-split filters or very large parameter ranges.
What is the difference between the original DS1, and the DS1MK3?
The DS1-MK3 adds quite a few features to the original DS1. The original DS1 started with a two DSP design and 48kHz maximum sampling frequency. So there were some compromises due to limited DSP power. The DS1-MK3 uses five DSPs and supports up to 96kHz.
Would you call it the work of your life?
No, there is more in the pipeline :)
What convinced you about Softube?
We had contact with Softube about since they were founded a long time ago. They have a very good reputation with their plugins. Their work with the DS1-MK3 and the additional derivatives is excellent.
Did you know their plugins before?
Not really, as I am not a user.
I know that the original DS1 code has been ported to the software line by line, so in theory, it should be identical to the real thing in terms of performance.
Is the effect of your work really a hundred percent clone of the hardware device? In your opinion, is the plugin different from the original?
It won’t be identical as the floating point formats in the DSPs and the plugin CPUs are not the same and thus there will be very small differences.
Where do you think we should look for these differences? Where are they most notocible?
That is difficult to tell. We tested several settings between hardware and software DS1-MK3 versions, but it is close to impossible to test all possible settings. In general one can say that floating point word-lengths play a role when a large and a small number are added. The small number gets dwarfed by the error the large number may have.
Softube has increased the processing resolution from 24-bit/96kHz to 32-bit/192kHz (it’s still 40-bit internally).
Did it translate into a difference in sound and behavior of the hardware plugin?
I don’t think the sound is much different. There are differences as mentioned above, but very small ones.
What exactly gives the work in 40 bits? What are the advantages?
The SHARC DSPs we use in the DS1-MK3 are 32 bit floating point and for some operations they can be used at 40 bit floating point. I.e. the mantissa is 32 bits instead of 24 bits which results in a better signal to noise ratio.
What else have you add or changed with the team of Softube?
With the plugin there are two additional types of safety limiter sections. And the POW-R dithering is not included in the plugin.
A good idea was to "split" the device into several stand alone plugins (MM-1 Mastering Maximizer, Deesser, and Compressor/Limiter) that are easier to use (simpler interface) and less CPU-intensive. Was it your idea or the developers from Softube?
That was the idea of Softube. I think it is a good move. The DS1-MK3 takes some time to be mastered.
The DS1-MK3 does not have an auto-makeup gain mode, it has a button for “Max” gain makeup, which essentially normalizes the output of the compressor to 0dB. How it calculates this in real time?
This is the auto gain makeup, actually. It can be calculated out of the parameters threshold, ratio etc.
The WEISS DS1 MK3 plugin Has a very high-quality dithering processor however, this is something that was useful more in the era of 16-bit PCM 1630, DAT, and CD recorders. For the most of today's users these feature could be not necessary...
Is dithering still relevant and useful in the era of streaming today?
Depends. If the streaming is done in a compressed format (MP3 etc.) then the encoding is preferably done from a high-res (e.g. 24 bit) file. Dithering to 24 bits is not really necessary in my opinion. But there are streams at 44.1/16 FLAC (e.g. Tidal). There it is good to have a source properly dithered to 16 bits.
As you mentioned, you worked as an engineer for Studer Revox, so I can't ask for your opinion about software Tape emulations. What do you think about all this tape emulations plugins available on the market? What do you, as a person who knows the tapes, think about them? Do you think it is possible to fully reproduce the correlation between the recorder, head, tape and audio tracks?
I don’t have any experience with tape emulations. In our high-end HiFi line we did an emulation of Vinyl playback.
All I can say is that with analog electronics and mechanical systems things can get very complex. It is a matter of how far one would like to go regarding emualtion. There are so many sources of non-linearities or added noises / tones or frequency modulations in such systems – it is crazy. “Intelligent compromises” are necessary.
Tell me something about your Ibis console? It is also based on the BW102 system - about which we were talking about earlier?
Yes, as the bw102 system is modular we kept adding modules and ended at building digital mixing consoles.
So, It's a digital console, isn't it?
How would you describe its sound to someone who never used it?
As it was based on the bw102 system the whole configuration was determined by the modules added. Simple channel strips were just input level, two AUX sends, channel fader, group fader. I.e. very transparent sounding which was the design goal. Other channels had EQ and Dynamics added, as necessary. Sony Music in New York used a 24 channel IBIS with GML fader automation for the mixdown of classical music. The AUXes were too few, so for pop productions the IBIS was not suited.
Since we are in the digital domain...
Do you have never been tempted to create your own software mastering workstation? A kind of dedicated mastering DAW based on your algorithms ? It would be something unusual... a complete professional mastering environment created by Daniel Weiss himself…
Maybe we should have done that – but it is a huge task.
Daniel, I can't waste this opportunity to ask you about one more product, a product that I know you're going to launch soon... I'm talking about speaker with built-in crosstalk cancellation... It's a very interesting idea...
Yes, we call it the Livebox. The basic Crosstalk Cancelling (XTC) principle is explained on that website (www.livebox.audio). The goal is to generate a more live-like experience for the listener. The effect of the XTC is that the instruments become very well positioned on the stage and that the acoustics of where the recording has taken place is extremely well reproduced. It is a “being there” experience. Some recordings work better than others for that method of playback. Best are binaural recordings (dummy-head, Jecklin Disc, Ambisonics rendered to binaural etc.) which usually are listened to via headphones for perfect crosstalk cancellation. The headphones have their own problems though (in-head localization), so it makes sense to play back via XTC speakers.
I know that you and your company are also active in the high end Hifi sector.... Would you like to tell me something about the devices in this sector? What makes them stand out from the competition? How do you intend to win this market?
The highend HiFi market is very crowded with lots of manufacturers. We try to do products which are not “me too” products in order to be competitive. For instance in the latest D/A Converter and D/D Converter we included a DSP which offers algorithms, some of them completely new to the audiophile clientele. Like De-Essing, Crosstalk Cancellation, Vinyl simulation, room EQ, creative EQ, constant volume keeper, loudness, crossfeed for headphones etc. I try to convince audiophiles that it is absolutely fine to enhance the audio signal, e.g. for better room compatibility or for matching their taste. The idea of not touching the artwork of the engineers who produced the recording is wrong. Because the artwork is touched as soon as it is played in a particular system / room. So I better take care that it sounds the way I like it.
If you were to describe the hardware DS1 MK3 from the user perspective, how would you describe it? What are its strong points, what is its advantage over other mastering compressors, de-essers?
It is a single band compressor / limiter. The band-split filter can be set to a wide range of frequencies allowing the use as a de-esser but also to treat just bass frequencies for instance. The band-split filter is linear phase for utmost transparency. It has a unique auto-release function. Other features: Parallel compression, sidechain stereo linking, preview delay, peak / RMS detection, overall delay can be set to a fixed value. POW-R dithering. 128 snapshots.
A few words for those who are still hesitating to try the Softube WEISS DS1-MK3 plugin?
I suggest to forget about all the comments etc. and just give it an unbiased try with the 20 day trial period. The DS1-MK3 plugin also includes the other three plug-ins mentioned above.
Thank you very much Daniel, for the meeting and for all the information. It was an extraordinary experience to talk to you!
I thank you, Adrian. It's been a pleasure too. Greetings to all the readers of Studio Knowledge Magazine!
Site Map About
Copyright © 2017-2019 studioknowmag.com. All rights reserved. The content on this web site may not be reproduced or distributed in any form and in any manner in any of the fields of exploitation, including copying, photocopying, and digitizing, without the written permission of the Owner.
All product names, company names, band names and trademarks are the property of their respective owners, which are in no way associated or affiliated with studioknowmag.com.