Exclusive Interview With Daniel Weiss.

I am doing fine, thanks. Business is fine, provided we can get all the chips we need. We are also hit by the chip shortage to some extent.

Compared to 2008 in high-end HiFi audio the most dramatic change is the advent of the streaming platforms, which even are capable to do hires sources. The high-end HiFi community has accepted that change in music distribution, as it turned out that the audio quality is by no means lowered by streaming instead of playing files from a hard disk or playing a CD. 
In the near future we probably see 3D audio – but it has yet to be seen in what form it will come to the consumers. Currently it comes in binaural form for headphone playback. At a fairly questionable quality, unfortunately.  Many companies are working on that topic.

In our high-end HiFi product line we got the Series 5 units which are the DAC501, DAC502, DAC501-4ch, DAC502-4ch, HELIOS D/A Converters as well as the DSP501, DSP502 D/D Converters. Then there is the MAN301 network player and the INT204 USB to AES/EBU Interface. Plus the PSU102 Power Supply and the Chiron cable family. And the SARACON sampling frequency converter software. Recently we added the DAC204 and DAC205 D/A Converters which are lower priced units.

Only the INT204 and the DAC204, DAC205, SARACON, PSU102 we sell in both product families. The pro audio family in addition includes the EQ1 parametric equalizer, the DS1 compressor / de-esser, the DNA1 denoiser / declicker, the DAC1 D/A Converter, the ADC2 A/D Converter, the A1 mic preamp / de-esser.

These are our latest D/A Converter designs. They are the same except for the size and for the 4 pin headphone socket the DAC502 has in addition to the ¼ inch Jack socket. Both are Roon Ready. The DSP algorithms, though, make them absolutely unique when it comes to high-end HiFi signal chains. The philosophy behind the algorithms is that we are convinced that the listener should be able to help him / herself when a recording is not quite sounding right or when the room is acoustically problematic or the speakers can use some more bass etc. 

So currently the following algorithms are available:
- For speaker signals: Creative Equalizer (a tone control), De-Esser (to get rid of sibilance sounds), Vinyl Sound Simulator (to get that special sound), Room Equalizer (to eliminate room modes), Loudness Equalizer (to listen at low levels), Dynamics (to get a constant volume for night time listening), Crosstalk-Cancelling (to get a never heard listening sensation).
- For headphone signals: Creative Equalizer, De-Esser, Vinyl Sound Simulator, Headphone Equalizer (to equalize certain headphone models with a predefined EQ setting), Loudness Equalizer, Dynamics, Crossfeed (to get a better headphone playback for tracks initially produced for speaker playback).



Recently we released the -4ch versions of the DAC501 / DAC502. Those currently have the same features as the standard DAC501 / DAC502 but the hardware is capable to play four independent channels. With the necessary software which we will release, the four channels can be used for various interesting playback setups. There is a white paper on our website describing what we have in mind with the four channels. https://weiss.ch/wp-content/uploads/2022/07/White-paper-on-the-DAC50x-4ch.pdf

The DSP501 / DSP502 have the same DSP features as described above, but have digital instead of analog outputs. So they can be used to enhance a D/A Converter with DSP algorithms and with the Roon Ready feature. Or they can be used to feed speakers with digital inputs directly. Such speakers will be seen more and more as speaker manufacturers nowadays tend to turn to active speakers.



The Medus we had to discontinued recently as we do not get the displays anymore. We just released a new flagship model, the HELIOS. It is based on the DAC502 platform with a new analog section. 

The main progress in DSP for audio is in the processing power available per chip. This means that we can run more complex algorithms than before and/or run them at higher sampling frequencies. Or run different categories of algorithms which simply were not possible to do (economically) with slower DSPs. Machine Learning and Artificial Intelligence have their first applications in audio processing.

If we are talking about running the same algorithm on a DSP (“hardware”) and a computer (“software”) then we would not expect any difference. Only the processor limitations in terms of word-length can cause a difference. DSP often are 32 bit floating point while computer processors often are 64 bit floating point. 

So there may be some differences. Of course we can think of algorithms where 32 bit floating point vs 64 bit floating point make a big difference (e.g. adding small numbers to large numbers) – so in essence it depends on the algorithm we are looking at.


If you are talking about the software emulation of analog hardware then we talk about a completely different thing. Analog can be very complex when it comes to the mechanics of non-linearities. And to model that in software may be extremely demanding, maybe even close to impossible….

Provided the source files are the same and the streaming service does not change anything audio – wise then the two methods are equal in quality.

If they do not mess with the audio they are absolutely fine. The main difference between players nowadays is the ergonomics with the user interface.

The DS1-MK3 is not suited for a high-end audio system. It is too complex to operate for laymen. It is as you would give the customer a violin to “make your own music”. In the Series 5 units described above we have the De-Esser algorithm which is one thing the DS1-MK3 can also do. The De-Esser in Series 5 has one parameter to operate it, i.e. more or less De-Essing. The art of engineering the Series 5 De-Esser was to tune it such that it works with just that single parameter. So we try to get some of our pro audio algorithms to the high-end audio market by considerably reducing the number of parameters in order to make them useable by laymen. This is not easy to do with as few compromises as possible.  

Our digital equalizers use an algorithm to emulate the basic characteristics of an analog equalizer. So they are fairly similar. The pro audio EQ1 in addition can be switched to linear phase mode which would not be possible in the analog domain. 

There are various EQs:
- The creative EQ is a 3 band parametric equalizer. Each band can be switched to one of the modes being low cut (eliminate low frequencies), high cut (eliminate high frequencies), low shelve (boost or cut low frequencies), high shelve (boost or cut high frequencies), peaking (a bell shaped curve for boosting / cutting at any frequency). “Parametric” means the frequencies, the amounts of boost / cut and the width of the bell can be set by the user. 

The creative EQ is an elaborate tone control.
- The Room Equalizer has 5 bands which can be set in frequency and cut amounts to suppress certain resonance frequencies. These frequencies (room modes) depend on the geometry of the listening space. For simple room shapes they can be calculated: https://amcoustics.com/tools/amroc?l=0&w=0&h=0&r60=0.6 


- The Loudness Equalizer is based on a Patent by Mr. Schwede. This equalizer is most useful when listening at moderate levels, e.g. during the night time. It makes the sound nicer. It does not use the commonly seen Fletcher-Munson curves, but some radically different curves with amazing results.


– The Headphone EQ allows to select headphone model specific curves to get a flatter frequency response from headphones. Currently many of the Audeze headphone models are supported. Other brands / models are in the works.

Some algorithms I already described above, like the various equalizers and the De-Esser. 
With the Vinyl Sound Simulator we try to emulate what a vinyl recording / playback does to the music, like dynamic range limiting, distortion, low frequency noise, limited dynamics at high frequencies etc.  


The Dynamics algorithm analyzes the loudness of a track and adjusts the volume such that all tracks played have about the same volume. This is useful when listening at night (at lower levels) or for background music playing. 
The Crossfeed algorithm kind of simulates the speaker playback via headphones. It generates crosstalk between left and right – similar to what you get when listening to speakers where you hear both speakers with both ears.
The Crosstalk Cancelling see the next question.

Crosstalk Cancellation (XTC) is a very interesting technique to achieve a “being there” sensation. 
Some time ago the “dummy head” was very popular for recording two channels. That is a plastic head with microphones in the ears, placed in the concert hall and recording the performance via those microphones. The playback happens via headphones to make sure the two ear signals are properly brought to the two ears again, i.e. without any crosstalk between left and right ear. Unfortunately it is not that simple, i.e. such dummy head recordings result in some immersive audio experience, but it is by no way close to a real “being there” experience. 

Similar to what we know when listening to “normal” stereo recordings via headphones – the sound location often is “in head” and not at all similar to a speaker based playback.



By making a “big headphone” with speakers, i.e. by taking care that the left speaker only reaches the left ear and the right speaker only reaches the right ear, one can enhance the immersive sensation considerably. This is possible with crosstalk cancelling which works similar to a noise cancelling headphone. The left speaker sends a cancelling signal which suppresses the signal coming from the right speaker to the left ear. And vice versa. This makes for a suppression of like 10 to 20 dB which is already enough for the system to work. (By the way, a very simple way to achieve that is to build a mechanical barrier between the speakers, i.e. have the L and R speakers close to each other and sit in front of them, on the center line. And have a large cardboard (or such) wall between the speakers extending up to your nose. Now your ears are kind of in separate rooms, each ear with its own speaker.)


When you do such an XTC setup the sound stage in front of you extends way beyond the speakers, to the left and right and also above. The speakers cannot be located anymore. And the room acoustics (reverberation etc.) at the location where the recording took place is reproduced in a very convincing manner. The musicians on the stage can be easily located. 
Our Series 5 units have an XTC algorithm built in. And we are soon releasing a soundbar which is an all-in-one system including crosstalk cancelation. See here: www.livebox.audio 

In terms of production / revenue the high-end audio is a larger part of our business. As for the engineering both parts, pro audio as well as high-end audio are equally demanding as we try to achieve top notch products in both markets. 

A friend of mine, who at that time was located in Australia, wanted to do a “no compromise” D/A Converter and we took the challenge. Of course “no compromise” is not possible in real life, but we used like e.g. resistors which cost 10 dollars a piece but had the advantage that you can specify any resistance value you need. And we ran 16 channels of D/A in parallel for a single audio channel. This to gain some better signal to noise ratio. My friend designed the frame and that was a bit over the top in terms size and weight… In any case, we never made a commercial product out of it. Today I would take another approach to a “no compromise” DAC I guess.

If the music has a soul, because it has been composed and played so well then you can do almost anything to it, you won’t be able to destroy that soul even by playing it through a mediocre setup. Don’t expect that playing music via a 100’000 dollar setup would reveal the soul if it wasn’t there to begin with… That is why playback via cell phone speakers works, still. But I can’t stand it if e.g. the bass is lacking…

Certainly helps to enjoy the music, but won’t bring out the soul.

There are recordings on CD (i.e. 44.1kHz / 16 bit) which sound better than many high resolution recordings. I tend to think that CD quality is enough, provided the musicians / engineers / producers do an excellent job. There may be genres of music where a higher resolution may be beneficial, so to be on the safe side 96kHz at 16 Bit is enough. YMMV.

That is still on our list, we even have a prototype. We’ll see. No promises.

I don’t think it would be of advantage, unless somebody would like to make all music as loud as possible…. A limiter can do a lot of damage. Our Dynamics control plugin I mentioned is kind of a limiter with a specific job to do. But basically music should be listened to without any change in dynamics. Some recordings (e.g. of classical orchestras) can be very demanding in terms of dynamic range and thus are not very well suited for playback in noisy environments.

One has to be aware that proper room correction starts with acoustical measures which take care of the main room problems like strong room modes (resonances at low frequencies). In residential rooms it is often difficult to install what would be necessary, e.g. large bass traps, though. So the next step can be a room equalizer and/or active bass traps like the AVAA from PSI Audio. Active bass traps are small and effective and thus are an alternative to large passive panels. The room EQ can take care of the remaining  problems like not too prominent room modes or high frequency reflections at ceilings or windows. With all those means of treating acoustics a living room can be made into a nice listening room. The main issue is that listeners often are not aware of the problems their room has. Education is key. I think room acoustics are slowly getting into the mind of listeners. A good thing.

If a room is “extremely well-treated” then it won’t have any (extensive) room modes anymore. But yes, room modes can be very annoying. The first step to get rid of them is to find out whether there are any in that particular room. This can be found out with a test signal we supply with some of our products. A listener can be very used to the acoustics of his/her room and thus may think that all is fine with the room. But in fact there are e.g. boosted bass frequencies which in fact are detrimental to the music.

I guess that is our EQ1 and DS1 units in the pro audio (equalizer and dynamics control) and the various D/A Converters we did for pro audio as well as high-end HiFi.

Our goal always is to design no “me too” equipment. Like for instance our DAC501 / DAC502 D/A Converters which have those plugins I mentioned. You won’t find that with any other product. And for customers who like to dive into enhancing the sonic sensation of their setup these processing features are very interesting.

We are working on various audio frontiers – The 3D audio setup is becoming popular in pro audio. But how can we bring that to the consumer? That is very demanding if you are not ready to install like 14 speakers in your home. Ideally we could have a 3D audio sensation via a 2 channel playback. This is what Apple Music is trying with Dolby Atmos 3D audio rendered to headphone playback. A very compromised 3D sensation, unfortunately. Many researchers are working on that topic, though.


Another frontier we are testing the waters of is audio for gaming. This is kind of a similar problem like the 3D audio in that currently game audio oftentimes is made in a 7.1 format, but most users have headphones as their audio system and not a 7.1 speaker setup. We try to achieve a compromise between those two extremes – with great results so far, I think.

I think that audio reproduction still has a long way to go. Despite the very high fidelity we achieve in the playback. Fidelity is one thing but the other important thing is to get the “right” signals to the ears. Right in the sense that we would like to achieve a “being there in the concert hall” sensation when listening. I think that we have not really achieved yet. 
In the meantime we can enjoy the music playback as we know – and love – it. •