thanx to Sebastian for translating it, Tik to review it, T.Ray to make the final corrections. you RULE GUYS!
and a special thanx goes out to Daniel Weiss from Weiss engineering (www.weiss.ch
) for spending some time with me and do the interview. the interview took place a few month ago in his researchlab in Switzerland.
How long is your average work day?
About 8 to 9 hours. Ah, the normal office routine. I just noticed that you seem to answer your e-mails around midnight.
Developers are usually known for their obsessive work practice. Every bit of spare time is devoted to developing new gear….
Music is a hobby of mine.
So you don’t think about your newest converters day and night?
No *laughs*, I wouldn’t say so. I see that you were formerly employed at Studer/Revox. What inspired you to start Weiss engineering?
Well, one day a Studer customer (from Harmonia-Mundi-Acoustica, a high end label for classical music) requested an interface to link his two digital units.. (Sony F1 and 1610, older video based 16 bit recorders). These were built around proprietary formats and interfaces. Studer did not have anything that could address the problem, as they did not make any custom gear, so I began working on the request in my spare time. This particular customer owned a mastering studio. This was 1984, a time that saw the emergence of the CD as a viable format. In those days there existed a large a gap in the digital audio processing market. The NEVE DTC (Digital Mastering Console), one of the few “serious” available mastering tools, was too cost prohibitive, and there wasn’t anything else that met the needs of the market. Meeting these needs became the catalyst for my new project. The customer would tell me what was lacking, and I assumed responsibility for the technical implementation. The company grew from there, and over time the customer assumed the responsibility of sales manager. That is how it all began. Why does modern music sound “worse” than older music? The gear has gotten much better!
Is it really worse?
Do you mean music-wise or from the technical point of view? Both! On the engineering side things like over-compression seem to have become the norm. Technical proficiency seems to have gotten worse too.
Well, we can’t do much about the music. If we look from the technical point of view, the fact that we CAN make it loud may be to blame. There are many things that seem to be worse, but certainly not everything. I think that it is going to be better again, and that people are generally becoming more and more aware of these problems. This is especially evident on some of the audio forums. Quite often blame is laid at the feet of the producers, but on the other hand, the engineers could take it upon themselves to say “we must resist that”. In the end I cannot say for sure what is right or wrong; some people can resist it, others cannot.
The big players, like Bob Ludwig or Ted Jenson, are famous for mastering loud, but I don’t really know if they must do so or want to do so. It is probably more a question of what each different musical style requires. It also seems that modern acoustic jazz recordings are compressed to hell and back. That is strange. Let’s talk about converters:
We find ourselves asking the same question over and over... Some say that “the higher the sampling rate the better”; However, Dan Lavry points out in his publications that a sine wave can be represented perfectly with 44.,1 kHz and 24 bit word length.
As long as it is limited to half the sampling frequency, yes... Exactly, Nyquist.
My question is: Should everybody record with 384 kHz?
No, I agree with Dan Lavry that 192 kHz is exaggerated. However, It IS important to note that the Nyquist theorem is true. That is absolutely clear. A classic example would be of the analog filters you need to limit the signal range. These analog filters would have to attenuate from 0 to 100 dB between 20kHz and 22 kHz .That is a very steep filter, and as you know, leads to phase distortion. It is it beneficial to sample with 96 kHz in the sense that the analog filter can be much flatter. You set the audio band to 20 kHz. So, with 20 kHz, you can set the transition from 0 to – 100 dB at between 20kHz and 48 kHz. This results in a shallower curve decay and, as a result, a nicer phase run. Going back to 44.1 kHz would require conversion into the new sampling frequency, and the re-implementation of the steep filter with a digital filter. One added bonus of working in the digital domain is that this entire process can be linear phase.
Bob Steward (Meridian) is more radical in his thinking, and suggests 60 kHz as the ideal sampling frequency. That is a quite tangible opinion, and one that I agree with. So does this mean that you are already working on your first 60 kHz converter?. :-)
No *grins*, it is not an accepted standard. If digital audio were created today,60 kHz would be a very good baseline. CD evolved from video. From an audio standpoint this format is not suitable. Would you say that recording at a sampling rate of 96 kHz is correct?
Yes. Are you saying that working with 44.1 kHz is fundamentally wrong?
In principle, no. This depends entirely on how the conversion is implemented. The implementation in the AD converter is really the most important factor. There are no doubt substantial differences in the quality of the various filters being used. In the past, people tended to make everything as good as possible technically, E.g. by a implementing a very steep attenuation beginning at 20 kHz. Doing this gives outstanding measurement results, and of course looks very good on paper. However, from a practical point of view, this is not necessary, as generally, the higher the sampling frequency, the lower the audio levels tend to be. In most cases it is safe to assume that at high sampling frequencies, the threshold is at most 20-30 dB beneath the maximum level. This difference can then be subtracted from the necessary attenuation. This results in more unobtrusive filters.
In the past, Apogee offered analog filter modules for A/D and D/A converters. The “G” version did not have a steep filter edge. These components were renowned for their quality of sound.
Is it true that, in the process of down sampling a 48 kHz recording to 44.1 kHz, one loses more than was gained by recording at 48kHz to begin with?
I would say that today we have sampling rate converters that are very transparent at both 44.1 and 48 kHz. It isn’t such a big deal. If for some reason you see an advantage in recording at 48 kHz, then by all means do it. But it has to be noted that when downsampling 48 to 44.1 one loses the potential advantage which a 48khz recording could have, namely a less steep anti-aliasing filter. Is it possible that some converters are designed for optimum results at. say for 44.1 kHz, or 48 kHz , and that better results may be gleaned from working at these “target” sampling rates?
It could be in that the analog or digital anti-aliasing filter is designed such that the slope is more gentle at 48kHz, because there is more room for it between 20kHz and half the sampling rate in the 48kHz case. So it could be that a converter sounds better at 48kHz. I doubt that one would see a converter sounding better at 44.1kHz. Speaking of converters and over sampling in general: What exactly does the term “oversampling” mean?
Today’s converters do eight times oversampling and more. These converters generally do not quantize very high, e.g. 5 bit with 64 times oversampling.
If I use my RME converters for 24 bit/44.1 kHz conversion, what sort of converter module is involved? Is it a 5 bit module that does 64 times oversampling?
Could be. One really cannot tell in general. In the past there were 1 bit Delta/Sigma converters, which did 64 times or 128 times oversampling. These were sampled down with appropriate noise-shaping. “24 bit” only tells us the amplitude of the word length at the converter output. This says nothing of the quality of the conversion. Some read 24 bit and think that it is TRULY 24 bits! This is not the case. Very good converters do well to have 20 bits of resolution!
So is this “standard” just a clever marketing strategy?
Working with a 24 bit word is certainly good for signal processing. The AES/EBU format is 24 bit. So it has been established as a “standard”. However; there is no real reason, from a practical sense, to implement a 24 bit converter module.
Is the whole story with the headroom (-144 dB with theoretical 24 bit) invalid?
I do hear a difference between 16 bit and 24 bit in the sequencer…
Well, digital signal processing is different. We must make a distinction between the analog and digital domains. Digital processing (e.g. gain) produces bigger word lengths. Tell me why I should buy a WEISS D/A converter.
*laughs* The only rule is that you should buy it if you like it! I would never say that our products are better than any others. It is entirely up to the end user to decide.
What makes a D/A converter “good”?
It must be transparent, unaffected by the cables that are plugged into the analog outputs; must have enough output level, should have negligible jitter-sensitivity at the input, and of course should have various desireable features. Of course signal to noise and distortion ratios must be first rate.
I feel it is very important to discuss jitter/output impedances. Converters often have relatively high output impedances. D/A converters, for example, should have a very low impedance so as not to be affected by cabling. Some think it desirable to have a tube in the output stage. Tubes, by the very merits of their design, are not transparent. I think for mastering tasks, transparency is an absolute must. You do not produce, mix, or master yourself. How do you identify with customers and their preferences?
As long as one is in constant contact with the customers, and continuously seeking feedback, no problems should exist. We depend on our customers to tell us what is or is not working for them. Our manufacturing methods are such that the technical standards and measurement data are always excellent. Often this technical excellence translates to great results on the audio side of the spectrum. Have you ever called Bob Ludwig to arrange a quick evaluation session?
*laughs* No, it is not like that. We finish the gear and then send it to interested persons. Those persons tell us what is good, and what is not so good.
What’s the secret of your de-esser? Does it have anything to do with the vinyl acceleration limiter? (Question by Jules)
No. The quality is based on two principles: one of which is the linear phase filter, where the entire band remains untouched, and the other is side chain oversampling. When the de-esser is idle, it is completely bit-transparent, that is to say if it does not perform gain reduction, it is on bypass. These two principles are why one does not “hear” our de-esser. Will you ever produce a plugin?
Saracon is not a plugin, if you mean that. It is native stand-alone software.
We are currently developing something like that. Not software, but an external DSP box. It is already making rounds online. It is compatible with MAC and PC systems, and has firewire, Ethernet, and AES/EBU connections.
The advantage of our concept is that all of the computing power is entirely isolated from the workstation. This means no sharing with other manufacturers or the Operating System. The copy protection is quite secure, *grins*, and the computing power is fully scale-able. The signal processors are implemented as modules. In our first version we have ten signal processors. These can be changed and renewed whenever one chooses to do so. To gain more computing power, one simply purchases more modules. One Gearslutz user has stated that your soft limiting is the only one that is really useful. Why do you make soft limiting better than others?
These days we are working with different concepts. My current way of thinking is that the converter should not be set to maximum level. With “24 bit” converters it is not necessary to utilize the A/D converter to the last 1/10 dB. It is beneficial to set it so that the peaks never get into the red zone and then raise the level in the digital domain afterwards. A side effect to this is that a soft clip function is no longer needed.. For example maximum -6dBu?
One must set the level conservatively. What problems can arise when a converter is pushed to the limit?
When one approaches 0 dBFS (maximum digital level), the distortion grows. This can be a good or bad thing, depending on which converter is used. Seen from that vantage point, the soft clip function would be an advantage. Could one also use it as a security blanket to prevent clipping?
Of course. There will always be a reason to use something for a different purpose than which it was intended. *smiles*; but yes, you are correct.
We developed the soft clip more as a safety circuit for the A/D converter. Clipping can cause very, very ugly things to happen *laughs* these ugly things were why we created the soft clip function! What is the best sample rate for recording?
I guess you are alluding to the 48 kHz vs. 96 kHz discussion from earlier. Many would say that it is easier to scale down from 88.2 kHz to 44.1 kHz than from 96 kHz to 44.1 kHz. It does take more effort to scale down from 96 kHz to 44.1 kHz; as bigger filters are needed.
*A sample rate converter is basically a low-pass filter, with the respective management of the coefficient of the filter(digital filters).*
When converting from 88.2 kHz to 44.1 kHz, one must use a low-pass filter that separates at 22.05 kHz. When converting from 96 kHz to 44.1 kHz, there are more filter coefficients, and, as a result, more memory is required. That is a disadvantage, but not a huge one in and of itself. The conversion from 96 kHz to 44.1 kHz can absolutely be as good as one that converts from 88.2 to 44.1 kHz. A bit more resources are required, but in these days, that is no problem.
There can be sound differences among various converters, depending on the sampling rate used, because, as stated before, the sample rate is converted at the input, by the implementation of filters. These filters vary from brand to brand. With the various filters come artifacts, and different sound results. What is your favorite piece of gear?:-)
I like synthesizers, especially the software thing on my computer (NI Reaktor). With that I can build much more complex instruments than with those analog synths in the early days.
I also like the old mechanical devices, tape recorders…. I still own an old Revox reel-to-reel tape recorder, but it is of course obsolete today. And my favourite right now is the Marimba Lumina, an incredible instrument made by Don Buchla.
When talking about EQs, we hear a lot about linear phase... But I really dig the MassivePassive and other pieces partly because of the phase shift.
You do offer gear with linear phase options, yes?
…Yes, we also have this as an option. It is a new option for sound designing, not necessarily a “better” one. It totally depends on the application. One cannot say in general terms that you should only use linear phase in mastering. An EQ is a device to shape the sound (this is in stark contrast to the purpose of a D/A converter). A MassivePassive is a very good mastering tool! It all depends on what one wants to achieve, and on the context……. What are the arguments for linear phase, and where it is best applied???
Users say that it is very transparent. Bob Katz owns one of my EQs and said it was “very clean”; and for some applications too clinical and transparent. What do you think the future has in store for the “hardware vs. software” debate? Or, in other words, how long will it take before computers begin to have “quasi unlimited” computing power?
That is probably not too far off anymore. In a few years we will have Teraflop per second performance in a PC. But the tendency is such that: The more CPU power is available, the more complex the algorithms that are put in place. When creating plugins, there will always be compromises. They should require little CPU power AND sound good. This is why we enjoy our “Powerhouse” so much; we don’t have to share computing power with other manufacturers or worry about hitting an upgrade plateau. We can upgrade as much as we want! We can design our algorithms the way that we choose, without worrying about compromise... (as much) Many people want to have hardware nowadays. It looks good, and doesn’t lose value as fast as software. It is always available, doesn’t crash , and so on. Are there any special anecdotes from your career that you want to share with our slutz?
When I presented the first EQ (Gambit) at an AES in Denmarks, we had to plug it to a computer to make it work….Well, first we had to buy a computer from Copenhagen. We went through a tremendous effort to buy it, and then someone (yes, me) spilled a beer on it!! It all worked out well eventually….. Do you like your job?
Mostly, yes, but at times, certainly not... When are those “not” times?
Well, I am the director, and sometimes I have to cut through the vast “paper jungle”. That is not really my favorite past time…. I heard that you play bass in a band. Are there any releases that we should check out?
No, you are quite lucky that we do not have any releases. *laughs* What are your goals for 2007?
Various new gear. The “powerhouse”, for example. Of course we would like to extend our reach further into the “High End” sector. Why is your gear so expensive?
There are several reasons. Production in Switzerland is very expensive (everything is made in Switzerland, from the circuit board to the case). Design and development is very expensive, too. It is the age old chicken-and-egg problem. We only have a few products, and because of that, our products cost more. If our prices were lower, we would definitely sell a greater volume, but we would have to find another place to produce, which of course would mean that we would end up in China or another low cost country. Eventually we would end up with something like Behringer. It is a vicious cycle!
And then you would have problems figuring out what to copy…..
*laughs* I would rather not copy Mackie… What is Your favorite dish?
Zürigschnätzlets (Sliced veal or pork in a special “Zurich style” sauce).
My wife works in the film industry, so I have seen a lot of movies. I have recently gotten heavily into Anime. I really dig Princess Mononoke….
If your house was burning and you could only rescue one thing..…
...my cats. :D Thank you very much for the interview and all the best wishes for the future.
All the best for you, too, and send my virtual greetings to your slutz.