FEATURE

SUPER AUDIO CV

The Sydney Opera House Converted

What do the Sydney Symphony Orchestra, audiophile recording label Octavia and renowned conductor Vladimir Ashkenasi have in common? They’re all recording to DSD at the Sydney Opera House. Text: Andy Stewart

The Sydney Symphony Orchestra is furiously rosining its bows in preparation for an onslaught of recordings the likes of which the Sydney Opera House has never seen. In the next three years there are 35 albums slated for release on SACD of the SSO performing with the famous Russian conductor and renowned pianist, Vladimir Ashkenasi. Late last year saw the recording of the first group of these multi-tracked performances: several pieces by Rachmaninoff (whom Ashkenasi personally worked with as a young man). With 2009 looming large on the SSO’s calendar as the year Ashkenasi will take up residence in Sydney and become the orchestra’s chief conductor, preparations for the ‘Rachmaninoff Sessions’ have been highly charged, and for the Opera House Studio, it’s also included the leap of faith into the world of DSD. For anyone unaware of who Vladimir Ashkenasi is, he’s like Elvis Presley turning up to your singing lessons, or Mohammad Ali teaching you how to box. A graduate of the Moscow Conservatory, Ashkenasi has recorded prodigiously – over five decades – everything from the Preludes and Fugues of Shostakovich to Alexander Scriabin’s sonatas, Rachmaninoff’s, Chopin’s and Schumann’s entire works for piano, Beethoven’s piano sonatas, as well as the piano concertos of Mozart, Beethoven, Béla Bartók and Sergei Prokofiev. From 1987 to 1994 he was the principal conductor of the Royal Philharmonic Orchestra in London, and from 1998 to 2003 the chief conductor for the Czech Philharmonic. He’s also won five Grammys over the course of his illustrious career. In short, he’s a monster on piano and no slouch with a baton.

AT 46

For Tony David Cray, who engineered the Rachmaninoff sessions from the Opera House Studio, recording Ashkenasi and the SSO was a real treat. It was also an eye opener recording to DSD (Direct Stream Digital) and working with the Japanese Audiophile label, Octavia, for whom these hi-res recordings were taking place. As Tony remarked almost immediately during our conversation in the control room: “Ashkenasi was awesome. He’s knocking on 70 but acts like he’s 15! His appointment is nothing short of incredible. He’s a beautifully natured guy and amazing with the orchestra – which bodes incredibly well for the Sydney Symphony next year. And for this first round of recordings – performances of Rachmaninoff – it’s comforting to know that Ashkenasi is renowned worldwide as the Rachmaninoff guru.” The conductor’s appointment is clearly manna from heaven. The recording deal between Octavia and the SSO (35 albums in three years is more than many full-time recording studios could boast) will not only see the Opera House Studio catapulted onto the world stage as a recording venue, it will also help remove the stigma the SSO and the Opera House management have long attached to recording sessions – that studio mics visible to a live audience are a no-no. In short, the old ‘you can record the orchestra so long as no-one can see the mics’ approach has been swept away by Ashkenasi’s appointment, and for Tony this means recording can now finally take place in the main hall in front of a live audience without these unreasonable ‘visual’ constraints. The combination of flying Decca Trees, new

microphones, a state-of-the-art Direct Stream Digital path and a world famous conductor means that the recordings have taken a giant leap forward, although in Tony’s words: “there are still a lot of ‘unknowns’”. I started the conversation by asking Tony how the three parties – the Sydney Symphony, Ashkenasi and Octavia – came together in Sydney. VLAD THE CONDUCTOR

Tony David Cray: When Ashkenasi came out to work with the Sydney Symphony late last year, part of the deal was that he bring his recording team with him. He’s actually signed to Decca but also has a very strong relationship with the Japanese record company, Octavia, who have been recording all over the world for years. Octavia are very well respected in both the Super Audio CD and orchestral recording markets, so when they asked us to do the recordings we jumped at the chance, and quickly set about tooling up to record to DSD, which was their preferred format. We’d never had a massive budget for recordings nor was our level of enterprise anything like that of Octavia. We’d been recording the Sydney Symphony on a comparatively modest level and we certainly weren’t recording to DSD. We were tracking through Euphonix A/Ds to typical PCM formats at sample rates of either 44.1 or 96k (which sounded fine), and most of the albums were being recorded directly to our Pyramix system. So the Sydney Opera House management committed to getting the equipment together and eventually we settled on a 24-channel DSD Pyramix setup. I was hoping for 32 channels, but the money involved was just too crazy, so 24 it was. But at that stage Octavia were usually doing 16-track orchestra recordings, so initially 24 seemed fine. Andy Stewart: Is 16 tracks a fairly typical number for an orchestral recording? TDC: It depends on the approach, the orchestra and the hall. Ideally you need a minimum of about seven, although you can do some fantastic recordings with just two! Lots of microphones generally only come into play when you’ve got individually strange things going on: an unusual percussion section or featured soloists perhaps… or if things aren’t going well! That’s when you have to get in close and start balancing the orchestra. Usually I find miking up an orchestra works out to around 20 channels. AS: What attracted you to the Pyramix option? TDC: I spent quite some time investigating which manufacturers could record DSD files: Genex and Pyramix systems were the obvious candidates. I also came across a Sony system called Sonoma, but finding out more about it was like trying to find the Lost Ark. It was also very expensive, so I quickly formed the view that since we already had a good working relationship with [Merging] Pyramix, and since it was also the program Octavia used, we’d be wise to stick with what we knew. We then discovered that Pyramix’s version 5 software max’s out at 16 channels of DSD.

Then, just as I was trying to find ways around that problem by potentially running two systems sync’ed together, the guys at Pyramix rang and told me they would have version 6 of their software ready to go by the time we were due to record. Version 6 could do as many as 32 channels of simultaneous DSD recording – so they claimed – but, of course, by the time we were ready to go, their release date had been set back. In the end we ran the beta software of version 6 on a hired, take-no-prisoners PC, plugged in the two MADI leads and ‘boom’, off it went. There was more to it than that of course, but essentially the system worked from the moment we set it up. AS: What about converters and monitoring, did they cause any dramas? TDC: Well, the Euphonix can’t replay DSD files of course, being a PCM (Pulse Code Modulation) digital console, and the converters only go up to 96k anyway, so we had to look around for new ones. We settled on Danish Audio Denmark converters, which Merging re-brands as Sphinx 2 (eight-channel A/D and D/As), and they formed the head of our chain. That was all set up on stage in the main hall of the Opera House and the signals piped down here to the studio and pushed into the recorders. We then fed a converted digital signal out of the recorder into the Euphonix. What’s important to realise is that we were never actually hearing DSD, we were only ever listening to a PCMencoded 16-bit/44.1k version of the recorded DSD files… Curiously though, I’d still have people come in here and be bowled over by the sound. AS: So because the Euphonix architecture can only replay PCM digital formats, you had to convert the DSD files to monitor the sessions… and at 44.1k? TDC: Yep. However, in reality I had nothing to measure the quality of the DSD recordings against, in the sense that on the very first day of recording, we had new acoustic drapes in the hall, a new conductor, incredibly enthusiastic, happy musicians, $100,000 worth of new mics, new converters and a new recording format. Everything sounded different, not just the recording medium, so it wasn’t easy to tell what improvements were attributable to the DSD rig or, for instance, the new Neumann mics floating on the Decca Tree. From my point of view, that was the change that had the most significant impact on the sound – the Decca Tree. Convincing the SSO to allow me to hang a Decca Tree in front of the orchestra (which I’d never been given permission to do before!) was 80 to 90 percent of the sound, I’d reckon.

Microphones liberated: the Rachmaninoff sessions saw microphones take pride of place in ways never witnessed by a live audience at the Opera House.

AS: Is there a difference to be heard in the sound of the DSD recordings then, in reality? Presumably capturing the orchestra via DSD counts for something… TDC: It does, for sure. After about 11 albums with the SSO I started to hear what I could only describe as ‘grain’ on the PCM format. It was almost like I’d zoomed in too far on a photo. After five or six hours in a session I’d

www

www.octavia.co.jp

AT 47

start hearing this granular sound, this ‘fur’, and realise that I couldn’t push the files any further, given their resolution. Even at higher sample rates where the recordings were focused on the integrity of audio, I’d still eventually get to the point where I could sense this ‘grit’.

Vladimir Ahkenasi conducts the SSO.

Then when I heard DSD for the first time in the Tokyo offices of Octavia, I was immediately struck by its creamy tone, this smoothness. There wasn’t any grain, even from two rooms away! I thought to myself, ‘My god, that’s it… this is what everyone’s talking about!’ Cut forward to doing these recordings, and it’s definitely a smoother, more transparent sound.

“ We’ll chuck up a mic and

then make sure it sounds right, whereas the Japanese crew will bring out a laser measurer and write down the distance between the instrument and mic to four decimal places!

”

HI OCTAVE PERFORMANCE

AS: Tell me about the experience of working for a Japanese audiophile label? TDC: There were a few interesting cultural differences. As we know, Aussies are pretty relaxed; we’ll chuck up a mic and then make sure it sounds right, whereas the Japanese crew will bring out a laser measurer and write down the distance between the instrument and mic to four decimal places! (laughs) But this approach didn’t really make much sense for these sessions because we were often ‘striking’ the microphones four times a week to cater for something else that was happening at the Opera House. And you just know the mic’s never going back in the same place! AS: And I guess there’s the other small matter of musicians moving during a performance. TDC: Exactly. But it was Octavia’s session and that’s the way they wanted it done, so that was cool. But certainly most things, like the setup of Decca Tree and spot mics, were essentially the same. What was fantastic was that because Octavia was so highly regarded by Ashkenasi, the SSO wanted to do everything they could to please him, which meant everything we needed to do they’d let us do. In the past, there was no way I could hang a giant Decca Tree over

THE DECCA TREE Tony David Cray: The Decca Tree consisted of five Neumann TLM50s – they’re groovy new versions of the old M50s, which are a curious omni that actually have a slight directionality about them in the upper frequencies – the higher the frequency the more cardioid-like they become. There are three of them in the centre and two on the flanks… This aspect of the setup was by far the most significant aspect of the recording.

AT 48

the orchestra, for instance, and have leads hanging down onto the stage in front of a live audience. Cut to our first recording session with Ashkenasi and there’s a Decca Tree hovering four metres above his head, and between him and the audience a mic stand extended to suspend the leads – simply unheard of for the Opera House prior to that… and Ashkenasi was so relaxed he’d come up and hang his coat on it! Incredibly, when the SSO management looked at the setup they said; “Tony, finally it looks like a recording session!” So we’ve come a long way. Three years ago they wouldn’t even allow a mic to be visible from the audience, now the management here reckon a Decca Tree looks proper! AS: That’s the green light for you then, isn’t it? From now on you can put mics wherever you like… TDC: Well, I’m still pretty timid about mic placement. I don’t want to stick up 15 mics in front of an audience if I can avoid it. We try and do what we can and be clever with how we mic things. But certainly the Decca Tree in front of the orchestra is the money shot, and ever since we’ve been allowed to put that up the sound of our recordings has improved vastly… it’s a fantastic step forward. SPOT THE MICS

AS: Were there any interesting tricks Octavia had up their sleeve to record the orchestra? TDC: They used the trick of time delaying the spot mics to match the time arrival with the Decca Tree, which I was interested to see. We’ve also experimented with that in the past and I’m in two minds on the benefits of that approach. Sometimes it seems to create more problems than it solves… AS: What about their perspective on the ‘listening position’ of the recording? Any differences there? TDC: That’s a good question, one I asked myself countless times to make sure I didn’t

WORKING WITH OCTAVIA Tony David Cray: Octavia have a fantastic workflow method, they can churn through albums and they’ve got a brilliant way of producing stuff by simply getting performances onto tape rather than relying on post production. They’re relentless in making sure a piece is played properly – to me that was the nicest thing I took away from the Rachmaninoff recordings. Despite all this technology they were always fighting

for the need to do things thoroughly, even if it meant doing the session again. The orchestra and Ashkenasi might have consequently gotten a little grumpy now and then, but Octavia’s producer, Masato Takemura, had the mettle to say: ‘Well, we just don’t have it…’ And the end result is this great recording that you can cut together.

impose my prejudices on their recording session. When you listen to Octavia recordings the strings tend to be a little more present than other aspects of the orchestra. They prefer what I’d call a ‘hyper real’ perspective. To that end there were some interesting comical moments; at one stage I asked: “how does Tomo [Octavia’s main engineer who couldn’t attend the Opera House sessions because of illness] like to mic the piano?” and they’d say, “Okay, one minute…” At that point they’d contact Octavia in Japan via Skype and the result would be… “U87 flat… two degrees up…”. Two degrees up! (laughs)

“ The D112 on the floor

underneath the bass drum gave the front-end sound an amazing power when added to the room sound. It was just astonishing.

”

AS: Were there any more unusual mic placements used? TDC: One thing that was very effective was close-miking an orchestral bass drum with an AKG D112. I know it might seem obvious to a studio engineer, but it’s certainly not the typical orchestral approach. I’ve always recorded percussion from above, miking ‘areas’ as it were. The D112 on the floor underneath the bass drum gave the front-end sound an amazing power when added to the room sound. It was just astonishing. Funnily enough, it was exactly how you might mic up a kick drum. They also had this fantastic Neumann GFM 132 hemisphere boundary mic placed in front of the basses on the floor. The mic records a half-omni pattern, given that it’s on the floor, and it sounded fabulous. I’ve always used a large diaphragm condenser chest height in front of the double basses with a top-end roll-off to capture more bottom-end energy and help push 40 and 80 cycles. The hemisphere mic did the same thing with no effort. PUSHING THE LIMITS

The irony of these initial Rachmaninoff recording sessions was that by jumping to a hi-res format like DSD, Tony and fellow engineers Jason Blackwell and Todd Dealy, were limited, not just by sheer track count, but also in their working methods. It was almost like a return to a two-inch tape mentality. TDC: There were all sorts of limitations AT 50

recording to DSD. The main one was that we wanted to do a surround mix, but couldn’t because there weren’t enough channels. For workflow purposes I also like to do a live [stereo] two-mix for reference to the multitrack as well – having each session primed with a two-mix makes workflows much easier. We generated 30 days of recording on umpteen hard drives – gigabytes of recordings being generated every day – and it would have been nice to be able to record a stereo reference mix back onto the files as we went. You don’t want to have to pull up a mix for every one of those sessions. But we couldn’t because we’d run out of channels. DSD OR DXD?

There are two hi-res recording formats in Pyramix, which confuses people no end, and the nomenclature – DSD and DXD – doesn’t help matters. DSD is to SACD what PCM is to CD. It’s a single bit stream recording method where every sample is simply ascribed a value, ‘0’ or ‘1’. These values are relative values, meaning that they represent the change in amplitude of an audio signal compared to the sample preceding it. So once you cut it, this relative value is lost. To solve this problem and allow the files to be edited, Pyramix uses its own DXD format (a standard – albeit hi-res – PCM format) that can be cut, mixed and processed etc. I spoke to Tony about the distinction and how they interact with one another in the real world. AS: Can you explain how DXD interacts with DSD inside Pyramix? TDC: Unlike DXD, DSD has to be approached like a 24-track tape machine… quite literally. It does nothing but record ‘in’ and replay ‘out’, that’s all you can do. You can ‘crash’ an edit like a blade cuts tape, but you can’t do anything else, you can’t change a level in DSD, nothing at all. AS: So there’s no mixer… TDC: No. AS: No editor? TDC: No. Well, you can make as many cuts as you like, of course, but there are no crossfades.

PYRAMIX & THE OPERA HOUSE

Tony David Cray: My experience here of doing a 50/50 mix of rock/pop and orchestral editing is that, for orchestral editing, Pyramix is God’s gift. It’s incredible. It’s so much fun. It abstracts you from the task you’re doing and lets you be creative. For instance, the way you can group things together and collapse them is fantastic. In your ‘arrange’ window you can have multiple timelines so, for instance, right now I’m just about

over the place. It just feels like I’m working with the performance because all the technical aspects have been abstracted away from me, which is awesome. When I have to do cuts with people in ProTools it’s far slower and very ‘linear’. But then, as a tape machine and DSP engine ProTools is fantastic. There are some big differences between them though, that’s for sure.

All you can do to a DSD file is cut it. You can’t do any level changes or processing whatsoever.

enhances the product to any great extent, from the consumer’s perspective, is very hard to say.

AS: So for these sessions you were literally recording to the DSD recorder like it was a two-inch master tape?

I think it’s a bit like surround sound mixing; yes it’s definitely worth the effort and it’s certainly a step forward but what is the rollout of it, what will be the end result be? That’s much harder to predict.

TDC: Exactly. We didn’t make any cuts to any of the multitrack files, nor generate any two-channel mixes. What we often found ourselves wondering was whether the files we were generating would be right for the mixing process back in Japan. It was a bit of a guessing game on many levels. AS: So unless you have an analogue mixing console, presumably you need to work in DXD, to mix it, since you can’t ‘mix’ DSD files together inside a computer. Is that right? TDC: Yep, that’s it. DXD is a Pyramix format that converts DSD to 352.8kHz/32-bit floating point files. The curious thing is that the data rate of DXD is three times that of DSD. It’s very high end but extremely heavy on the juice as a recording format, so it’s simpler and less DSP intensive to record DSD files initially and convert them to DXD for mixing later. AS: And is that what Octavia is doing now with the DSD files back in Tokyo? TDC: Yes. They don’t have an analogue console so they’re mixing from DXD because the only way to mix DSD together is in the analogue domain.

AT 52

to edit some recordings I did recently with the SSO – four performances of around 30 tracks each – 120 tracks. With Pyramix I can collapse them down so I’m only looking at four things. What’s more, I can click on that ‘single’ track and then solo those 30 channels, route them automatically out the same bus, edit them and crossfade them etc. I’m actually handling 120 tracks in the session, cutting and flying multitrack audio all

HIGH REVELATION FORMATS

AS: You mentioned the track count limitations of the format, but did the ‘creaminess’, as you describe it, of the DSD format encourage you to record things differently or say to yourself, ‘Wow, for the first time I’ll be able to put a mic here and have it count for something!’? TDC: DSD is definitely something you’d pull out when you had a pristine pianist or a fantastic vocalist and acoustic guitarist. But personally I’d be recording to DXD; I can’t see the point in going to DSD and then converting it to DXD just to get the job done. On a purely pragmatic level working with DSD is still difficult. For instance, currently the DSD/DXD Pyramix formats won’t allow us to use the Euphonix console, although that’s predicted to change soon. The new Euphonix Supercore has been designed to cater for sample rates up to DXD so I should be able to use that sometime in the near future.

You have to remember that DSD was largely borne out of Sony/Philips wanting to produce a really cheap converter chip. To do PCM is expensive and complex; DSD on the other hand is simple, simple, simple and really cheap to make. It was never intended to be a multitrack format.

That’s why it’s been important for us to say ‘yes’ to the Japanese, because now we can offer clients the opportunity of recording to high resolution digital, whether it be DSD or DXD, and I’m hoping some actually take it up. We’re now in the privileged position of being able to produce a catalogue of Australian orchestral recordings that are of supreme quality. We don’t have many SACD recordings of orchestral works here, but hopefully that’s now set to change!

AS: Who’s idea then was it to record these orchestral albums to DSD?

AS: So do you stand at the door after concerts and sell punters SACDs?

TDC: Octavia’s. They love the technology. It’s also driven by market differentiation, of course, and the SACD market. We have 24 channels of high-end conversion and 24 channels of DSD recording capacity now and that gives us a technical advantage over most other recording setups in the country. DSD definitely sounds better – it’s clean and pristine and some of the performances on the recent recordings sound fantastic – though whether the medium really

TDC: Well, funny you should say that… I was just talking to the SSO and they’re interested in releasing some albums this year on vinyl as well as iTunes. What a contrast that will be!