The History of BIS' departure from DSD recording to 24 Bit 44.1kHz PCM recording.

Here for posterity's sake is the article I spiked at the request of the president of BIS, Robert von Bahr.  In June of 2009, I agreed not to publish it in Positive Feedback Online.  However I think the subject matter is of extreme importance to the future of SACD and am posting it here at SACD Lives.  

I have read this article dozens of times and Robert's request to spike the Positive Feedback Online is unfounded as much of the article are unedited quotes from him, his recording staff and other well known engineers.  It is mostly my comments he disagrees with, personally I believe I have been more than fair, indeed I bent over backwards to present his side of the issues.  I hope you enjoy the article and learn from all the issues brought forth.

The History of BIS' departure from DSD recording to 24 Bit 44.1kHz PCM recording.

I fully support BIS releasing 24 Bit 44.1kHz PCM SACDs as I firmly believe all recordings should be released as SACD/CD hybrids without regard to resolution.  I even believe old mono 78's should be released as SACD/CD hybrids, indeed in my opinion single-layer CDs should be a thing of the distant past. 

However I do believe in full disclosure, not only of the recording type, but in the case of PCM the sampling rate and bit depth of the recording and any mixing.  Not all companies that record PCM are as straight forward with recording information as BIS is, indeed there are many SACDs that I not only suspect of being from 44.1kHz PCM but 16 Bit as well, based solely on their sound quality.   

Since BIS quit recording DSD in 2005 due to the limited availability of DSD editing tools, the editing possibilities of DSD have greatly increased but Robert von Bahr has informed us that buying new DSD or even higher resolution PCM equipment is not viable in the current economic climate.  Indeed BIS is only in a position to currently offer 24 Bit 44.1kHz PCM Surround Sound SACDs and 16 Bit Bit 44.1kHz PCM Stereo CDs.

The nine known DSD recorded SACDs BIS released are generally excellent sounding and give a window into how good their engineering can sound at true high resolution.

Time Table
1) BIS switched from DSD back to PCM recording in 2005
2) Robert announced in 2005 BIS is producing everything from now on in PCM (24bit, 88.2kHz or 44.1kHz) I understood that to be 88.2kHz PCM for SACDs and 44.1kHz for CDs, I was wrong.
3) Robert announced in 2009 BIS is producing everything in 24 Bit, 44.1kHz PCM. Evidently at some point 88.2kHz was either abandoned or never used.

From a previous PFO article "The SACD format needs a more complex SPARS-type code"    24 Bit 44.1kHz PCM was not even listed as I was unaware of anyone actually using this combination of bit depth and sampling frequency.  Until BIS' announcement I thought 48kHz was the lowest sampling frequency available for 24 Bit.  As we all know 16 Bit 44.1kHz PCM is the resolution for redbook CD. 

BIS DSD SACDs versus BIS PCM SACDs in 2 channel stereo

So with BIS' latest SACDs what one gets is an increase to 24 Bit and multichannel sound.  As a 2 channel stereo lover I cannot address the multichannel aspect however I do hear definite improvement afforded by the extra 8 bits of information.  Speaking of BIS recordings only, if one listens to the first three DSD recorded SACDs from BIS' Grieg series and compare them to the last three PCM recorded ones the differences are very telling.  In my system the DSD recorded ones have smoother sting tone, more air and ambiance, warmer bass and more delicacy in the high percussion instruments.  The first three Grieg SACDs sound quite like what I hear in live music, the last three sound like recordings.  I found the PCM Complete Peer Gynt dry sounding with a distinctly PCM sonic signature, which I complained about before the 2005 announcement of the switch by BIS from DSD to PCM.  The Holberg Suite actually has very strident string tone which sounds like redbook CD and something I thought was impossible on SACD even from low resolution. 

1. Grieg: Norwegian Dances, Op. 35, Symphonic Dances, Op. 64, Lyric Suite, Op. 54 - Ole Kristian Ruud (conductor) Bergen Philharmonic Orchestra - BIS SACD-1291 -DSD Recording - Stereo/Multichannel Hybrid SACD
2. Grieg: Piano Concerto, Symphony in C minor, In Autumn - Noriko Ogawa (piano) Ole Kristian Ruud (conductor) Bergen Philharmonic Orchestra - BIS SACD-1191 - DSD Recording - Stereo/Multichannel Hybrid SACD
3. Grieg: Sigurd Jorsalfar Op. 22, Landkjenning Op. 31, Bergliot Op. 42, Sorgemarsj over Rikard Nordraak EG117, Den Bergtekne Op. 32 - Hakan Hagegard, Gorild Mauseth, Ole Kristian Ruud (conductor) Bergen Philharmonic Orchestra - BIS SACD-1391 - DSD Recording - Stereo/Multichannel Hybrid SACD
4. Grieg: Peer Gynt Op. 23 - Hakan Hagegard, Marita Solberg, Ingebjorg Kosmo, Kari Postma, Ole Kristian Ruud (conductor) Bergen Philharmonic Orchestra - BIS-SACD-1441/42 (2 discs) - 24 Bit 44.1kHz PCM Recording - Stereo/Multichannel Hybrid SACD
5. Grieg: Holberg Suite Op. 40, Two Elegiac Melodies Op. 34, Two Melodies Op. 53, Two Nordic Melodies Op. 63, Two Lyric Pieces Op. 68 - Ole Kristian Ruud (conductor) Bergen Philharmonic Orchestra - BIS-SACD-1491 - 24 Bit 44.1kHz PCM Recording - Stereo/Multichannel Hybrid SACD
6. Grieg: Peer Gynt Suites Nos. 1 & 2, Funeral March in Memory of Rikard Nordraak, Old Norwegian Melody with Variations Op. 51, Bell Ringing Op. 54 No. 6 - Ole Kristian Ruud (conductor) Bergen Philharmonic Orchestra - BIS-SACD-1591 - 24 Bit 44.1kHz PCM Recording - Stereo/Multichannel Hybrid SACD

Here is my comparison review A Tale of two SACDs - DSD versus 24 Bit PCM

DSD, PCM and real life

To my ears DSD recording is ideally the most realistic and the most consistent with what I hear in the concert hall.  However PCM from 24 Bit 88.2kHz and higher can often sound superb, until all recordings are from DSD masters I do find I have to either make compromises or live with a very small collection.  Neither situation is ideal, so when I see a recording company moving away from DSD it troubles me greatly! 

Back in February of 2009 after selling over 150 SACDs due to unsatisfactory sound quality compared to DSD recorded SACDs, LPs or live music I proposed a SPARS type code for SACD.   Most of those I sold were from masters of undisclosed resolution and I suspect they are low resolution PCM (44.1kHz and 48kHz), thus my proposal for a SPARS type code for SACD.  I will admit it is a selfish attempt to not lose money on low resolution recordings residing on SACDs.   While I do believe all recordings without regard to resolution should be released as SACD/CD hybrids I also believe in correct labeling so those of us who don't care for the sound quality of low resolution recordings can instead choose those from real high resolution masters.

The magic words for me are "PURE DSD" and "DSD recorded"  I sometimes play DVD-Audio and high resolution downloads and I can tell you NO DVD-Audio even at 192kHz sounds anywhere near as good as a DSD recorded SACD. When I really want to sit down and deeply enjoy music I put on a DSD recorded SACD.  I prefer Telarc the most, but other labels also make excellent sounding DSD recordings.  Once you hear how really great DSD recorded SACDs are it is very hard to go back to any resolution of PCM.  DSD is what makes the SACD format so special. And BIS used to make some excellent sounding DSD recorded SACDs, hopefully they will again in the future.

In selecting new music to buy, a disc having SACD logo is just the first step, and I would like to know how it is recorded before I buy it.  There is a real pattern that has developed over the last 10 years, DSD recorded SACDs sound the best to me, with the exception of a few known bad ones. Most of the SACDs I have sold as unacceptable sonically were either PCM or of undisclosed resolution. This tells me it is far safer financially to buy DSD recorded SACDs, as most of those stay in my collection. Over 90% of my SACD collection is DSD, there can be no greater compliment to DSD and SACD than that.  My strategy is I look for a DSD recorded version first, resorting only to PCM if DSD is not available.   

Even though editing in DSD is still somewhat limited compared to PCM (a good thing as far as I am concerned, unfortunately since writing this, DSD can now be manipulated as much as PCM, damn), I recommend the entire recording industry move to DSD recording!  And make more natural recordings where everything is done prior to recording and eliminate fixing it in the mix.  I was really hoping SACD and DSD would move the world back to honest recording where we at home can hear the artists as they really are. This is how Telarc and others record in DSD, and besides the sound quality, the mostly "hands-off" requirement is the other thing that insures DSD recordings are usually also excellent  performances as well.

The BIS' October 28, 2005 announcement
  
October, 2005 when we first suspected BIS' newer SACDs of not being DSD recorded, based solely on how they sounded.   Zeus at sa-cd.net produced a letter saying BIS was now doing everything in PCM (24bit, 88.2Khz or 44.1Khz). I am sure all of us thought it would be 88.2kHz for the SACD and 44.1kHz for the CD.

• Email From BIS' Ingo Petry to Zeus in response GoldenRedux's question "I was under the impression that all of the BIS SACD's were DSD recordings, but only some of them are listed here as DSD. Are these listings correct?"

"We are producing everything from now on in PCM (24bit, 88.2Khz or 44.1Khz) first leaving everything in PCM-domain for postproduction (mixing, editing, reverb if necessary) and do the transfer to DSD at the last possible state!!! That gives optimum postproduction quality (and recording quality as well especially for multitrack recordings up to 32channels) without compromising in premixing on location!" and "We strongly believe after several blind listening tests that our solution is overall the technical most clean AND sonically most convincing one."

• My response to the news on October 29, 2005 

"Most BIS recordings on SACD have been from DSD Masters and have sounded superb especially the Greig Orchestral discs, the first 3 are DSD the Peer Gynt doesn't say what it is... the even later BIS SACDs do not say how the master was made and they do not sound as good as their earlier DSD SACDs and I could not put my finger on why. So now we know!  On the other hand Chandos who as recently went the other way from 96kHz 24 Bit PCM to DSD recordings...have been superb so far.  BIS please reconsider as your SACDs from DSD masters are among the very best in my collection..."

The BIS' January 19, 2009 announcement

Email dated 1/19/09 to RWetmore in response to his question "What PCM format do you use?":

"We record everything 44.1kHz 24 bit. Of course in post production we also use 32bit as a recording format when applying level changes.

About 5 years ago we decided to embrace the SACD as an Audio Format for mainly 3 reasons:
1. By far superior sound quality than CD no matter what High Resolution Format you choose.
2. the only existing (surviving) carrier for Surround Sound - this is really a huge benefit in our opinion
3. Hybrid disc allows backwards compatibility with ordinary CD Players. thus we produce only one carrier

In the beginning we had generous technical Support from SONY, who wanted to push the format. but the logistics did not work for the large number of SACD recordings we intended to do. And the machinery was far too complicated - especially when honoring the fact that we do all our chamber music recordings with only 1 person. Orchestras we usually record with a team of 2.

You could argue that recording in a higher sample rate would make sense. At least 88.2 kHz or 96, if not 352.8 like a few others do.

We did try to do that, but we had to discover a limiting factor in the necessary computer technology.

The main point is that some of the tools we need to create the best possible (and natural sounding) mix simply will not work at sample rates over 96 kHz. And at 96kHz they already do not have enough Inputs and Outputs, thus forcing us to make other compromises which will become far more audible than the difference between 44.1 kHz and 96kHz.

Last but not least: The CD layer is still 44.1kHz, so no converting is needed here. Since many (if not most) of the listeners still play the CD layer rather than SACD, we believe that they will at least benefit a little from NOT CONVERTING the recording. While those who listen to SACD layer really receive an excellent product. Converting to DSD is different from converting from 96kHz to 44, many CD players have already 1bit converters, they use similar technology, but the amount of DATA recorded on a CD (16bit) is still the limiting factor.

I could have given you a Yes/No answer, but I hope you will understand, that there are a lot of factors behind such a decision. It is not so much the recording format that matters (we are talking on a high level), but rather how you take care to capture sound by placing the microphones in the best possible way and then how you handle the recorded material.

If you look at our catalogue, you will see how many SACDs we have managed to release. They are all carefully produced, edited and engineered. We do like to use our own equipment, so we have to ship it around the world. The advantage is, that knowing our tools well keeps our focus on the music and the sound. Handling too complicated or unknown equipment will absolutely distract from the really important things.

Thank you for your interest. Since your questions have been quite to the point I am actually curious about your own position on this and how this may have influenced your opinion.

Kind regards,
Thore Brinkmann"

• Robert von Bahr confirmation of this email:

"I think Thore Brinkmann's mail says it all. It is an upfront mail telling honestly why we have taken the stance we have. We're using our knowledge of microphone placing and production values to enhance music, using the SACD possibilities to give an added sound experience in the process. I believe now, and I have always believed that expertise in WHERE and HOW to record, psychological insights in how to get the absolute best out of the artists ultimately are much more important for the end result than what system one is using. And his comment

"The main point is that some of the tools we need to create the best possible (and natural sounding) mix simply will not work at sample rates over 96 kHz. And at 96kHz they already do not have enough Inputs and Outputs, thus forcing us to make other compromises which will become far more audible than the difference between 44.1 kHz and 96kHz."

Is a crucial one, because it exemplifies the difference in attitude between those who listen for the music's sake and those who listen with only numbers in their minds. Yes, we could record in 88,2/96 or higher, but, doing that, we would miss out on other things that "are far more audible". With such choices we opt to go with the music every time.

In the end of the day, it is music that we are producing, not bits."
  

Robert von Bahr explaining the switch from DSD to PCM recording

"I would like to be a missionary for OTHER production values than perceived sound quality differences, these other production values being so incredibly much more important to the end result.  I will even go as far as to say that EVEN if our sound could be marginally better, using pure DSD (I don't admit to it, so don't quote me wrongly), but EVEN IF, and the price would be to lose out on all the possibilities we now have to make the record the way the artists would like them, I would buy the lesser sound. It would be a musical vs. a technical war, and, for me, the MUSIC WINS EVERY TIME...I don't think the same musical results could have been achieved with a pure DSD recording.

I have so far conducted several tests with Golden Ears, great artists and what not, and not even the artists can differentiate between origins, even when they listen to their own recordings on their own instruments. 

If we were to go for pure DSD and thereby lose so many possibilities to make the musical contents of the recording so much better.

I absolutely respect those who think differently and who think they can actually hear the differences and who think that the sound is much more important than the music.

For those, to whom MUSIC is the most important thing on an SACD, they will presumably at some point stop buying pure DSD-SACD's, since inevitably the musical results are being negatively influenced by the non-existent possibilities to do any meaningful post-production within the DSD system, or then the artists play so safe so as not to make mistakes that the result becomes dull.  [Editor's Note: I highlighted this quote in red as I find it INSULTING not only to the engineers who make wonderful DSD recorded SACDs, but to high resolution digital and all owners of SACD players!]  We've been there, seen it, and drawn the consequences, because we asked the artists...And then we have those in between, who simply enjoy good engineering, good acoustics, and artists who produce to their absolute limits, and they'll enjoy what we and other musical companies do."

• My post about the DSD to PCM switch and Robert's response

I said: "I understand your reasons for switching from DSD to PCM as DSD is hard to work in as Michael Bishop of Telarc explains. Once the volume level of the recording is set it cannot be changed in DSD, the engineer must get this correct before the start of recording. Balancing, etc has to be done before recording, in other words everything has to be correct and perfect before you turn the DSD recorder on as there is no fixing it later. Electronic splicing is kept to a minimum as each edit point has to be converted to DSD wide or PCM for less than 2ms and then converted back to DSD. As Michael Bishop explains this is seamless and one never hears the recording leaving and returning to DSD as the edit point is so short, much shorter than a splice on an analog recorder. Most engineer's once they start recording DSD never revert back to PCM because they are so impressed with the results. Which makes BIS somewhat of an enigma being the only company I know of to adopt DSD and revert to PCM."   

   

Robert said: "Dear Teresa, and here you have it in a nutshell. I have earlier stated that the listening experience is a conglomerate of several factors, where the sound certainly is one. What hall you use is another. Musicians and instruments a third and fourth. They are all important!! We consider our sound to be absolutely first-class, and, so far basically everyone have agreed, with a few exceptions, you and raffells included, that is, agreed until they READ about what they have been listening to....

But here you're talking about another factor. The fact that DSD doesn't allow any meaningful post-production is VERY detrimental to the final results. [Editor's Note: This is not true any longer - see DSD in Production]  If I may use an analogy - how do you think that Rembrandt or Titian would have reacted, if someone simply took away their smallest brushes and said to then that they would have to make do with the big ones only? How would they paint hairs? And this, even if they have access to the best paints in the world?

For us, it would be TOTALLY frustrating not to be able to fix that cracked horn entry or correct other mistakes, balancing or otherwise, because the system is more like a piassava brush that you sweep streets with compared to hair brushes. You talk about 2 ms edit points. That would be the equivalent of a vertical edit with a pair of scissors (yes, I am that old). Let me tell you that many edits go on for seconds. You cannot edit in a lying tone "vertically" without there being a huge bump. We record in multitrack so as to be able to bring out the ultimate musical results afterwards. That's basically incompatible with DSD. [Editor's Note: Crossfade edit points are where sections of DSD material are electronically joined together, it is this point lasting for 2ms that is in PCM wide and as Michael Bishop explains is not audible.  The scissor cut is a great analogy, think of replacing a section of an analog tape, you remove the section of tape from the bad take and replace it with another section from a good take.  Your splicer will join the replacement section at both ends, thus there will be two cuts and the splicing tapes applied.  So in DSD that would be two crossfade edits one at the beginning of the of take and one at the end.  The electronic cut or crossfade edit point would be the PCM wide and where that actual music is would be DSD.  The edit points are only for joining the music which can be any length, as there is a crossfade edit point on both sides of the edit. (see next section for Michael Bishop's posts on crossfade edit points)]

But don't let's get technical here. We are talking about a system that really does sound well - no two ways about that.  But we are into MUSIC, and we want to present the music and playing in the best possible manner. That REQUIRES that we can work on the material we have recorded, and DSD makes that either incredibly cumbersome or downright impossible, unless you leave the DSD domain and revert to PCM - and then, what's the point, from your point of view? We just drew the inevitable conclusion and left the system altogether. Honestly, DSD should only be used for live performances, not carefully and lovingly produced records, IMHO.

As I said - as long as one has reached a certain level, we go for the most important things. We find our sound level to be so good that even experienced professional listeners cannot with certainty distinguish between the alternatives but we would totally deplore to throw away our chances to correct and choose the really best parts of the recording takes - and that's what DSD unfortunately does - see Michael Bishop above. Jared Sacks elsewhere has also testified to the difficulties in working with DSD.

The 2ms (or 0.002 second) Crossfade edit points 
     

Robert von Bahr does not agree with Michael Bishop that the edit point (where to music segments are joined together) can only be 2ms, he says "Please take it from me, you can only edit in 2 ms cross-fade in absolute silence.  Real edits occur in up to 5-6 SECONDS crossfade (especially so in organ music).  Most edits are in 50 - 1'000 ms crossfade, also with the "pure DSD" companies."  I agree with Michael Bishop's statements.

My information on Crossfade edit points are directly from Michael Bishop of Telarc and he disagrees completely with you!  He says the 2ms (or 0.002 second) edit point is too fast to be audible to the human ear.  Also that is the only time it is in high resolution PCM at the 2ms edit point on both sides of the edit.  The music itself is always DSD. 

• Posted September 19, 2003

"There CAN be (almost) pure DSD editing on the Sonoma DSD workstation, as long as the source is DSD, and that one does not change volume level of the source, the DSD stays as DSD except at the edit points. Only for the duration of the edit crossfade (as short as 2ms) does the DSD signal go through the Sony E-Chip for editing processing.

I try to keep all edit processing to an absolute minimum because of the interstep needed for the edit. Unity gain is the rule on my projects!

DSD workstations that rely on 32-bit pcm at 392kHz are not very successful at maintaining the source audio quality, IMO. The DSD source goes through the 392kHz process and stays there, regardless of edits, until the final conversion to 1-bit DSD."

Best Regards,
Michael Bishop 
Recording Engineer

The Light goes on It sort of goes like this:
DSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSD(2ms X-fade)DSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSD(2ms X-fade)DSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSDDSD etc., etc., etc.....

"if one can hear the millisecond conversion to extremely high-resolution pcm in a once-in-a-great-while well-placed edit, that person should be the subject of scientific investigation. A good edit is a good edit whatever the domain. It's all smoke and mirrors anyway." 


Best Regards,
Michael Bishop 
Recording Engineer

The problems Robert see with editing in the DSD domain

"We have different approaches. I come from the musical side and you from the sounding side. Both are absolutely valid, and we'll have to agree to disagree. 

I agree with what Michael Bishop says... If you want to restrict yourself to
- not change volume level and
- only edit, where the music allows short crossfades (meaning pauses),
yes, this is OK.

But, Teresa, we don't want to restrict ourselves thusly. We come from two different camps. Let me be specific, just one example of hundreds:

Take Stravinsky's Petrouchka for solo piano. There is a passage, where the pianist has to, repeatedly and in a prestissimo tempo, hit clean E octaves with both hands, wide apart in the descant and bass, and, between these chords, play in the middle of the keyboard. So, in effect, the pianist throws both hands repeatedly to the extreme sides and tries to hit those E:s cleanly. Take it from me: it is IMPOSSIBLE for any human being to do this cleanly and with any certainty, if the tempo is as ferociously fast as Stravinsky demands. Impossible!! So you have two choices - either to play safe, meaning slowing down the tempo, so that you can wiggle your head about and check where the hands are, or just keep the tempo and hope for the best, looking at either the left or the right hand. In the concert hall almost everyone chooses the latter option, fast playing, since the music is the important thing. But comes the recording. No one likes to present a record, where wrong notes, split notes are part of it. And this is then the problem: do they slow down the tempo, so as to play cleanly or do we use the editing tool to give the poor bugger several chances, and then choose the ones, where he hits it right, in the correct tempo? To me the answer is obvious. You state that: "This Direct-to-Disc which cannot be edited or changed in any way proved to me most editing and messing with the sound after recording is totally unnecessary."

I respect your opinion, Teresa, but I cannot agree.  We prefer the artist to take all risks in order to present the optimal reading of the music, NOT to play safe! Editing makes this possible.

Given then that we are committed to editing: just so that the music isn't combined from small takes of a few bars or worse, we encourage the artists to play long stretches at a time, to retain the flow of the music, something akin to having several performances, from which we take the best parts. But, Teresa, artists aren't machines. They don't play the performances exactly alike every time around (mistakes apart). Thus, when it comes to snipping between the takes, it is more the rule than not that small level adjustments have to be made, to make the transition from one take to the next absolutely smooth and inaudible. This is IMPOSSIBLE in DSD. If we want to edit in the middle of the music - and we do - we have to use sometimes very long cross-fades. [Editor's Note: According to Michael Bishop (and other engineers using DSD) crossfades are only 2ms long and are on both sides of the edit, the newly inserted music is in the middle of the crossfade points which can be as long as needed.]

 This is not an option in DSD, because the cross-fades in PCM would, with Teresa's ears, be audible. (Not true section: The 2ms (or 0.002 second) Crossfade edit points, below) So, we're faced with the dilemma:

- Do we choose the musically best takes for the master or
- do we let the recording technique limit the choices for us? Or
- do we want to present a product with audible and unmusical edits?
[Editor's Note: Sorry Robert no dilemma, invite Michael Bishop to your recording session and he will show you how to do this in pure DSD.]

Again, for me the answer is self-evident, as it VERY DEFINITELY is for the ARTISTS. They would never allow a technical detail stand in the way for them to present the work as they want, NEVER!!! If they knew about the choices, that is...
I honestly do wonder, if the artists who have agreed to work in DSD have been given these choices, but I do not wonder what they would say, if they knew.

So, Teresa, from our standpoint we don't really have a choice. We come from the musical side. If only the developers of DSD had gone all the way, but they are technicians, not MUSICIANS with any experience of recording situations. This is such a pity, because let's agree finally on one thing: DSD is fantastic. Even though we consider what we deliver to be so good as to be practicably indistinguishable from it. But, for a real musician (and at BIS, we are) DSD simply is not practicable. Such a crying shame."  [Editor's Note: Telarc, Channel Classics, PentaTone and others prove that not only is DSD quite practical but sounds vastly superior to PCM at any resolution.] 

Robert von Bahr's final declaration, May 22, 2009 on sa-cd.net

"Dear Forum,

I think it is about time to make a final declaration from BIS regarding this discussion.
I don't pretend that I will be able to convince everyone or even anyone, but one thing I think I am entitled to, and that is to be treated with a modicum of respect. I have a track record (no pun intended) for honesty, and I will not further entertain posts from people who doubt my honesty or call me names. I am fully willing to discuss all other aspects, and I do admit that I may be wrong, but I don't cheat, period. For those who remain unconvinced, I have saved the best for the last, namely that you are in no way forced to buy any BIS products. As a quid pro quo I would want your possible criticism to dwell upon verifiable facts, not just conjecture. So, here goes, and I am sorry, if it gets long-winded.

1) We have never pretended, or tried to pass off, any BIS product for anything else than what it is.  On the contrary, I have repeatedly on this Forum and elsewhere declared that, with a few exceptions in the beginning, we record in PCM, only converting to DSD in the final stage.

2) Why do we record in PCM rather than DSD?
Because, for us, the possibilities of editing, mixing etc of the raw material are infinitely greater in PCM than in DSD. Directly said: we can do more to the recording to conform with the intentions of the artists in PCM than we can in DSD. We consider this to be of utmost importance for the enjoyment of the final product.

3) Why do we record in 44,1/24 rather than in higher resolution?

Because we face a bottleneck situation in several important aspects.

- Our equipment, although being state-of-the-art, cannot handle the quantity of recording tracks needed in higher resolution than 44,1, since we record in multi-track. Any change in this would require a very sizeable investment in more/bigger equipment both for the 4 movable recording equipments we have as for the editing suites as well as necessitating more personnel for every recording. We have been constantly modernizing our entire recording equipment, incl. cables, for our SACD recordings in order to improve our sound (Multitrack recording with remix Stereo and Surround). However, these investments, before committing to them, also had to undergo a consequential analysis - we couldn't invest in equipment that necessitated an extra person per recording or making the post-production impossibly complicated, since we then would be in the classical situation: operation a huge success, but the patient (SACD releases) died.

- We only have one editing/mixing plus two other editing-alone suites for multitrack. It is really time-consuming to reset the whole equipment for different recording formats.

- We do an enormous amount of multi-channel recordings and the run-through time in this studio has to be kept at a minimum (even so, each multi-channel recording takes a good extra week in the studio just for mixing, with at least 2 persons constantly present - wanna know the costs?).

- We face economic realities like anyone else. The SACD format is, at best, a niche format, which we are trying to support by being the world's 4th largest SACD label, and probably the 2nd (after Exton) in newly recorded products per year. To make the investment in a totally different architecture of our equipment above and, taking into consideration that we would have to be 3 persons instead of 2 for each orchestral recording with what that entails (salary, travel, hotel, per diem) would make the whole SACD project financially impossible, and therefore diminish the quantity of released SACD:s drastically. We are already taking a horrible beating, financially, because of our commitment to SACD, and I am OK with that, but I will not jeopardize BIS for SACD:s sake. My earlier thread "How many of you..." showed conclusively that very few, even on this Forum, would support SACD to the exclusion of anything else, and now that we are faced with a boycott, the incitement to continue to release SACD:s doesn't exactly grow.

4) So are we sacrificing the quality of our SACD products for the sake of quantity?

A very important question, and essentially the one that set this whole discussion off. Some of you say a resounding YES to that - we say NO. Let me argue with the YES-sayers.

- We currently employ, full-time (and the only record company in the world to do so), no less than 6 Tonmeisters from Detmold, Germany, all of whom have moved to Sweden to work for BIS. Detmold is without any comparison THE place for the education of producers and recording engineers in the world. It is a very extensive education in music and technique, the entry exam is gruelling, as are the studies. "My" 6 producers have been chosen individually, out of many dozens, after long time of apprentice work at BIS before being offered a permanent position.

- We also have a small circle of some further 5-6 Tonmeisters from that same place who, while electing not to move to the country of sexually liberated icebears walking the streets, nevertheless have been chosen, after the same apprenticeship, to be part of the "inner circle" of BIS trustees.

- The point being that, before we took the decision to move from native DSD to PCM, we conducted an extensive double-blind test on all these people, and some of our best artists as well, to see what, if any, difference they could hear between the DSD original and the PCM version of the same piece. The result was statistically totally insignificant, that is, the result could have been achieved by tossing a coin.

- Am I now to believe that precisely the professionals on both side of the microphone we chose for these tests are actually deaf? And that only people like Teresa on this Forum are equipped with golden ears? That I personally, at age almost 66, am past hearing the extreme HF noises I reluctantly accept, but my Tonmeisters and world-famous artists? No, I cannot accept that.

- So what about Teresa's ("DSD") Golden Ears and veracity? I am sorry to go ad hominem here, but I have been quite upset with some of her assertions on the Forum. This is not to destroy her, just give an example that there may be OTHER reasons than what you ACTUALLY hear that decide what you THINK you hear. For instance recording format, numbers, preconceptions. The following are all DSD:s quotes on the BIS threads.

Quote (April 5, 2009): "Sorry I got off topic there, all I wanted to say is Robert's recordings are great...".  [Editor's Note: This was a discussion about the musical qualities of BIS' SACDs not their sonic attributes here is the complete sentence:]  "Sorry I got off topic there, all I wanted to say is Robert's recordings are great and I love the off the beaten track repertoire."  Here is my post:

Quote (May 20): "I have sold many SACDs because they sounded like they were from 48kHz PCM even when they did not state so! And 44.1kHz is even worse. My Golden ears are fine, none of my BIS SACDs are from 44.1kHz PCM. You just have too little faith in Robert von Bahr, shame on you. FYI the only BIS SACDs I know of from 44.1kHz masters are all the ones in the Ultra Extended Playing Time ...."  [Editor's Note: This is true and I am working of an article of 150 recommended and not recommended SACDs, and really did not believe that anyone would make new recordings in 2009 at only 44.1kHz!  That still sounds bizarre to me!] 

Then after the horrible exposure of how we record:
Quote (May 21): "I still say these two SACDs sound like high resolution PCM:
American Spectrum - Llewellyn
Shilkret: Trombone Concerto etc. - Lindberg/Neschling
Any information on these two as they sound way too good to be 44.1kHz PCM."

Well, as everyone will have guessed by now: The 2 SACD's referred to above are indeed recorded in 44.1/24, just like the rest.  Thus I maintain that people are far too influenced with what they THINK they're hearing rather than what they actually are hearing. I again apologize to Teresa - it is a cheap shot, I admit, but one that, under the circumstances, unfortunately needs to be made. Listen with your ears, not with your eyes. What you hear, ACTUALLY HEAR, is the truth. All other things are inconsequential.

[Editor's Note: The fact that I like the sound of these two SACDs is a supreme compliment to BIS as I have never heard a 44.1kHz PCM recording in any format that sounded as good as these two!  Robert von Bahr's original revelation from 2005 stated that he was switching to PCM at 88.2 and 44.1kHz PCM.  Robert now says these are both 44.1kHz PCM,  At no time did I ever say either one sounded anywhere near as good as a DSD recording.  But I think they sound really great for PCM.   So you see Robert this accusation is untrue and does not invalidate any of my extremely realistic sounding DSD recordings.  Here is my latest thoughts on these two BIS SACDs from June 14, 2009 - I have re-listened to these two PCM BIS SACDs and I stand behind my comments even after Robert von Bahr confirmed both are only 44.1kHz PCM.  They do have a PCM signature but all PCM recorded SACDs to my ears have a particular PCM sound even at resolutions much higher than 44.1kHz. It does not have the air, ambiance, bass presence or huge soundstage of a well made DSD recording but it never expected it to.  24 Bit is definitely an improvement over 16 Bit and I will continue to purchase BIS SACDs of unusual and rare repertoire otherwise unavailable on SACD as I have been doing since BIS' switch back from DSD to PCM in 2005. Finally Robert I listen with my ears, with the lights out sitting in the sweet spot, I never listen with my eyes!]

5) What constitutes a remarkable listening experience and how does one achieve this?

There are several factors that contribute to a good recording. Choice of music, musicians, instruments, hall, microphones, placement of same, producer that is psychologically attuned to the artists, and, yes, recording equipment. In order to achieve the very high level of customer satisfaction, one cannot really scrimp on any of these, but, conversely, not any single one of these factors is solely responsible for the satisfaction of the listener. They all combine to give the enjoyment aimed for.

Now, it is my contention that we are dealing with music here, music that has to be presented in the best possible light. All other things are secondary to that.  It is further my contention that one should take pains to make sure that the weakest link in the chain should be strengthened first. Thus it is more important that a bad piano is replaced with a good one rather than that a good microphone is replaced with a very good one. It is more important that a harpsichord is in tune in a reasonable church than untuned in a good one. The list goes on.

Eventually one has ironed out the weakest links, making them stronger, until one has the prerequisites fulfilled to have the chance to make a really good recording. Call it a "minimum requirement" situation.

So, where does the difference DSD/PCM or even 44,1/24 vs 88,2/24 come in as far as weak links are concerned? Rather far down, I'm afraid. Actually, at the bottom. Since we haven't even between ourselves been able to ascertain - IN REAL TESTS - any statistically significant difference between them, and we have proven (above) that even the most vociferous Golden-Ear-people can be embarrassingly mistaken, we tend to believe that other factors are much more important for the result of the recording than the system used in this particular case, once you pass a certain level, and 44,1/24 is defined as high resolution. A change in the recording system as above may result in sound differences that even our experts cannot define with any certainty, meaning that the difference, if any, is so small as to shrink into meaninglessness, whereas the microphone placement, or recording in the wrong hall can utterly destroy the recording, regardless of recording equipment. It is my contention that you're barking up the wrong tree, concentrating on a really insignificant part of the chain (given a certain level), while totally disregarding the things that have an immediate and extremely noticeable effect on the result. And, so far NO ONE HAS SAID A WORD ABOUT SURROUND!!! Now, if that isn't a factor that in itself is incredibly more important than the perceived change in sound quality I don't know what is.

Add to this the fact that going over to 88,2/24 at this very moment in actual effect would make the whole production chain - for us - so expensive that we would have to stop releasing SACDs altogether, one is faced with the alternatives

- either embrace what we do, and have been reputed to do so well, also in this Forum or
- boycott further purchases of BIS material on account of a perceived lack of quality.

That choice is up to you.

Hopefully this will cap the discussion as far as BIS is concerned.

Best wishes - Robert

Response to Robert von Bahr about current DSD recording by Everett Porter of Polyhymnia

"There's a middle ground to be found here!

We do LOTS of productions in DSD. Postproduction possibilites are somewhat more limited than with PCM, but the situation is constantly improving. It's now possible to do vastly more in post with DSD than it was a couple of years ago (I was editing and mixing 24 tracks of DSD today). Working in DSD takes more powerful hardware, the latest software, and sometimes more patience, but it certainly can be done. All the tools that you need are available in high quality, along with many you don't (mixing, dynamics processing, EQ, reverb, restoration plugins).

At the same time there are times where we're working with PCM -- always with hi-res (88.2 or 96k) if it's intended for SACD release. This is primarily in live situations, where we're providing live radio and/or TV mixes in addition to the recording, and where setup and breakdown times are often extremely limited. Working in PCM is vastly easier for live broadcasts (and more reliable -- all current DSD mixers are PC based), and post is easier. Most importantly hi-res PCM sounds great when the right equipment is used.

Lots of great SACD's are being produced from all sorts of source material (also wonderful analogue tapes!). Different labels, producers, and engineers will choose different methods, as they always have, to attain what they're looking for. Some put more emphasis on the purest possible recording chain, others on having the most creative possibilities in post. Personally I think that hi-res recording is worth the extra effort -- and I don't feel limited in any way by the technical restrictions imposed hi-res."

Jared Sachs of Channel Classics reply to Chris' post on sa-cd.net
    

"It is really not my business to lecture others of which recording medium to use. I can fully understand the barriers for Robert if he is dealing with all his sets plus being compatible with their system back home. Not to forget all the postproduction capabilities which DSD just does not have. Maybe if the SACD was selling to a larger market, he might have invested in the equipment but now is certainly not the time.

My only comment to many labels who approach me is that at least record and store the data at the highest possible denominator not knowing what the future will bring by way of format to the consumer. Downsampling is always a better selling point than upsampling!

Bis makes great recordings and everyone on this forum always seems to agree so indeed the music comes first and you accept the lower sampling rate for what it is. then whether you or Bis can hear the difference in a blind listening test is not an issue."

Michael Bishop of Telarc on DSD editing

"There CAN be (almost) pure DSD editing on the Sonoma DSD workstation, as long as the source is DSD, and that one does not change volume level of the source, the DSD stays as DSD except at the edit points. Only for the duration of the edit crossfade (as short as 2ms) does the DSD signal go through the Sony E-Chip for editing processing. I try to keep all edit processing to an absolute minimum because of the interstep needed for the edit. Unity gain is the rule on my projects!"

Best Regards,
Michael Bishop
Recording Engineer

How much would it cost you to upgrade your equipment to fully handle 176.4khz/24bit PCM?

"It would be exceedingly expensive to change all sets of equipments going round the world plus all the editing suites and it would necessitate an extra person at every recording. We can afford neither or, as I have said earlier, we cannot afford to actually record with them AND release SACD's.

The extra selling because of SACD is miniscule. The correct solution, if our finances were the only important thing, would be to cease and desist. We would save more than half of the recording/editing costs, and the loss in customers would be far less - don't forget, we sell SACDs at the price of normal CDs and we pay copyrights and royalties like on normal CDs. Given the extra costs for the physical production of the SACD we earn much less per copy sold, and, since the vast majority of copies are sold to people that don't even know what SACD is, much less can play them - we take a loss also here compared to normal CD's.

There is not much in SACD, financially, but we consider the sound quality difference, plus the fact that we can serve surround, to be more important than our profits. So far..."

BIS' commitment to SACD

"We have so far been totally committed to the SACD cause, trying to enlarge the repertoire (while not forgetting the standard repertoire), and, for us, the more, the merrier. But we are a business, after all, and forced to play on the Industry's conditions. We cannot charge more for our services (since most people don't want to pay extra for something they cannot use) and so we have an upper limit as to how much every production may cost. Personnel costs much (travel, salary, per diem, hotel), and most productions cannot carry those extra costs.

Then there is another thing: artists WANT to have the SACD treatment, since it is perceived to be more prestigious. We want to give it to them - thus we have to keep the quantity up. But the artists do NOT want to give up on musical quality, and they can be exceedingly demanding, so we have to employ a system where we can serve them on that front as well, while not wasting more time than necessary.

It is so nice (I am not being sarcastic here) that so many persons involve themselves in the thought process on how to ru(i)n a record company, but here you must trust me - I am sitting with all the answers, as far as BIS is concerned.

Some may say that we should lower the quantity of recordings and go on a spending spree for the few remaining ones. I think not, not, if you are concerned about the future of SACD. I think that the more material there is available, the better for the format. That we would go bankrupt in such a process is another matter, since the extra sales of such recordings would not pay for the extra costs of recording/editing them. And, again, as long as WE can state with honesty that our recording quality is not AUDIBLY less good than the other systems, I am proud to follow the line that we have chosen. Please believe me, we have had endless internal discussions about this before deciding. It wasn't just a whim from yours truly, Robert.

Ingo Petry, the most senior of BIS' recording engineers/producers explains the BIS recording process

"Example of a typical BIS microphone setup for large orchestra:

Main mics: spaced omnis, for example TLM 50 or KM 130 (all Neumann), if also surround sometimes an additional Center mic mostly of the same type as the AB configuration.

Surround mics: could be anything from omnis to figure of eight, depending very much on the hall and the music.

Spot mics: KM 143s, 184s, TLM 149s and other members of the Neumann family depending on instruments and/or musical content and/or hall specifics.

Distances, angles etc. run within a certain frame but are adjusted to the acoustical environment one has to cope with :-)

This is of course only a 'shopping list' and if further interested can be cooked into a meal within a more private conversation but definitely not on this Forum!

Let me use this opportunity to say that real relevant questions like microphone placement and types etc. are so much more rewarding to discuss than the lately slightly overblown recording format dispute...

The reason that many people think that BIS has a rather high score in 'good sound' is simply that we at BIS agree on certain fundamental sound ideas. That includes also the importance of certain elements which may lead towards a good recording. Terms like 'openness, no coloration, impact, smoothness, personality, naturalness etc.' are all words we try to commit ourselves to.

And I must totally disagree with anybody who claims that just because a recording is not done all the way in the DSD domain, it cannot live up towards that goal. Just because one is believing in something very strongly doesn't make it more true...

Anyway, this discussion is an endless one and the positive thing I can still see with some of these strongly minded opinions within this Forum is that as long there is a discussion there is an interest and somebody who cares! And that is definitely the case for BIS and all its contributors!

At last some idea for people who are willing to look a bit behind the scenes before making all these assumptions and black-and-white comments. I strongly recommend to try to come to a recording session (any label not just BIS) and see first-hand what is it all about! For our label not so difficult because we do recordings in almost any possibe (or impossible) place on this planet.

See you there,
Ingo Petry,
Record producer, sound engineer, BIS Records"

Why I sold my 44.1kHz BIS SACDs (updated 1/27/2011)

Last year I sold my last two non-DSD BIS recorded SACDs because ultimately I could not live with the lack of deep back and the somewhat strident string tone.  

SACD is not about making recordings one can live with it is about making the best sounding recordings possible! If we only cared about the performance we could subscribe to a monthly service and download BIS MP3s for 25 cents a track.

For serious listening I want the highest resolution possible and for me that is DSD recorded SACDs. I only buy high resolution PCM recorded SACDs of music not available in DSD recorded SACDs. And that is why I don't have a lot of non-DSD BIS SACDs. 

I am among those who firmly believe no music should be recorded at any resolution lower than 88.2kHz and if I had my way everything would be recorded DSD. Which to me means more natural sounding recordings not only because of the higher resolution but also because the editing options are limited. [Editor's Note: Unfortunately this has changed DSD can now be edited and manipulated as much as PCM without leaving the DSD domain, damn!]  I have always preferred a less hands on recording technique, more like photographic recording. For example the Crystal Clear Direct to Disc recordings of the Boston Pops and the London Philharmonic, especially the wonderful Gould conducts Gould Direct to Disc LP.  His Spirituals for Orchestra is so real and spontaneous plus the resulting sound quality is among the best ever captured. This Direct-to-Disc which cannot be edited or changed in any way proved to me most editing and messing with the sound after recording is totally unnecessary. I have always preferred recording companies who took the natural approach.

If you listen at all to Jazz music please check out DMP's Direct to DSD SACDs with NO editing within a track or song. If there is a mess up or problem with a track the entire track is re-recorded. Tom Jung believes editing destroys the spontaneity of the music and that is why he never edits, his results prove he is right.

I really wish there was a way I could talk BIS into going back to DSD!  Go to your stereo, put on this DSD recorded BIS SACD Grieg: Norwegian Dances, Symphonic Dances, Lyric Suite turn out the lights and let the absolute beauty overtake you, and then put on one of BIS' later PCM recorded SACDs and here what is missing. I will continue to praise BIS' DSD recorded SACDs as I always have even if they never make another one. 

Even if it turns out a couple of BIS' 44.1kHz sound good that is still no justification for purchasing SACDs from 44.1kHz masters, I want the highest resolution possible not the lowest! I don't want just good or even great sound, I want sound that is closest to the realism I experience in a great concert hall. So far that has come only from DSD recorded SACDs.

If Telarc, PentaTone, Exton, Channel Classics or Chandos abandoned DSD recording such as BIS did back in 2005 I would be even more upset as these are my favorite recording companies and their superbly realistic DSD recordings are why I own an SACD player in the first place, otherwise I would still be playing LPs.

How about 24 Bit 44.1kHz surround sound downloads?

Robert said: "...please tell me how to get 24 bit surround onto a DSD CD. It would save us so much money, and I could finally exit this Forum and stop producing SACD:s..."

I said: "Robert you can offer 24 Bit 44.1kHz surround sound downloads. See the Boston Symphony Orchestra Digital Store  They offer 24 Bit 88.2kHz surround sound and stereo downloads, 44.1kHz downloads would have the advantage of half the download time.

Downloads will eventually completely replace CDs, so this might be a good answer for you. I don't know if HDTracks plans surround sound downloads in the future, perhaps this is something you could try on your website?"
        

Final thoughts

Whither you feel 24 Bit 44.1kHz PCM is high resolution or not please remember BIS SACD/CD hybrids are priced the same as single layer CDs!   That is like getting a free upgrade to 24 bit with optional multichannel sound.   Looked at in this light they are most definitely preferable over single-layer CDs.

Further reading:

SACD and ultrasonics

DSD in Production

A Tale of two SACDs - DSD versus 24 Bit PCM

Everything is important to make a great enjoyable recording.

SACD format needs a more complex SPARS cod 

BIS SACDs