Antonín DVOŘÁK (1841-1904)
Symphony No. 9 in E minor, Op. 95
“From the New World” (1893) [44:38]
Leoš JANÁČEK (1854-1928)
Taras Bulba, Rhapsody for Large Orchestra (1915-8) [23:20]
Vienna Symphony Orchestra (Dvořák), Berlin Philharmonic Orchestra (Janáček)/Jascha Horenstein
rec. Symphonia Studio, Vienna, 1952 (Dvořák), live broadcast: Usher Hall, Edinburgh, 1961 (Janáček). Mono
PRISTINE AUDIO PASC603 [68:41]
If I’m interested in a given CD, there’s usually a good reason; on those rare occasions when there isn’t, it’s usually because there are two good reasons. This disc is one such. Back in the 1960s, as a youth both callow and hard-up, my main desire was to increase the number of pieces of music in my record collection. To speed things up I operated a strict “one recording of each piece of music” policy. Although I aimed to get “good” recordings, this policy had a rider decreeing that I make do with a slightly less well-thought-of recording if it was appreciably cheaper. Of all the names accumulating in my youthful brain, that of Jascha Horenstein stood out, but not for the usual reasons. To me, he was something of a legend; I often heard of him, but somehow never found his name emblazoned on any LP covers. Not surprisingly, in 1971, I was tempted by glowing reports of Horenstein’s recording of Mahler’s Third Symphony; unfortunately, I already had a recording of that work.
Around 1961 Barbirolli, through his revelatory live performance (in Bradford) of the Fourth Symphony, had in no uncertain terms introduced me to Nielsen. When a much-lauded new recording of the Fifth appeared, soon after the aforementioned Mahler, not already owning a recording of the work I seized eagerly upon this one. Odd though it may seem, I kid you not that, such was my haste, the conductor’s name did not strike home until I had the LP in my grubby little mitts.
Oh, how unutterably glad I was that I’d resisted getting Bernstein’s recording; this Horenstein one knocked it into a cocked hat, partly – I’ll admit – on account of Alfred Dukes’s spectacular snare-drumming (a perhaps surprisingly rare attempt to comply fully with Nielsen’s explicit instruction), but mostly because Horenstein did just about everything right, by which I mean “exactly the way I like it”. Thereafter, unaccountably and sadly, his name seemed to slide out of my ken back into misty obscurity.
That’s one good reason; now here’s the other. One area in which digital audio truly transcends its analogue predecessor surely has to be the restoration of old recordings. After all, the best – nay, almost the only – thing that could be done by analogue technology was to transfer 78s to LPs. Digital technology has spawned many powerful tools that, in the right hands, can work restorative wonders (we’ll draw a discreet veil over what they can work in “the wrong hands”). What’s more, its potential is still nowhere near exhausted. The outstanding representative of this continuing development of the state of the art must surely be Andrew Rose’s Pristine Audio – and, I’m shocked to realise, this is the first Pristine product to have come my way.
I can’t think of a better place to start peeling away the layers of legend than with one of the most popular of all symphonies, Dvořák’s Ninth. The well over 200 different recordings include an unusually large proportion of good ones– and an unusually small proportion of utter dross. In spite of my youthful policy (which, being a Yorkshireman, in later life I’ve never quite shaken off), comparatively speaking I seem to have accumulated a fair pile, including LSO/Kertész (Decca/London 430046-2), Halle/Barbirolli (EMI Phoenixa CDM 763774-2), LPO/Macal (CD-CFP 9006), Brno PO/Waldhans (OCCDs CD1/2008), plus a couple of Slaithwaite PO performances that I recorded myself. Without exception I enjoy all of them although, under threat of torture (quite mild torture is more than sufficient), I’d admit that my softest spot is for Barbirolli who, it just so happens, is a rare exception in playing, as Dvořák directed, the first movement’s second subject at the same tempo as the first.
Horenstein is one of the many who don’t, but he still manages to differ from just about everybody else: having established a slower tempo for the second subject, he carries it right through the whole development section, restoring tempo primo only at the first subject recapitulation. On first hearing, as you might expect, this quite startled me. However, by the third auditioning, I found it nowhere near as alarming as I’d initially imagined. “You just got used to it,” I hear you say. Well, no – I think it was just me being a bit slow to get the point! And the “point” is, I believe, that the slower tempo lends a bit of extra weight to the brief but turbulent development, casting the shadow of a question-mark against the first subject’s flightier eagerness.
The main thing is that Horenstein (way back in 1952) was challenging “received wisdom” about this hugely familiar music. A touch more subtly than Haydn in his 94th. Symphony, in this first movement he was thereby craftily jolting us out of our complacency; and indeed, again like Haydn, Horenstein doesn’t need to repeat the trick, since he’s now got us listening more attentively to what he is doing.
And what does he do? Quite a lot; and behind it all seems to be an acute awareness of the music’s emotional “current”, expressed through touches, mostly delicate, on the various rudders – of such as tempo, dynamic, phrasing and instrumental detail.
In recordings, this last seems to be widely regarded as largely the province of modern audio technology and consequently something for which recording engineers get the credit (or the blame). Contrariwise, in live concerts, the conductor alone gets the credit (or the blame). Now, really, shouldn’t this latter also be generally the case in recordings? Certainly, in recordings that purport to convey the concert-hall experience, engineers should be credited in relation to how faithfully they capture the details elicited by the conductor – as opposed to those noted by the producer for bringing out in the “mix”. Unfortunately, listeners to (and reviewers of) recordings have no way of knowing how to apportion the credit. So, we have to guess. Here, my guess, based on the age of the recording and the fact that I can’t detect any symptoms of “spotlighting” or other “fader fiddling”, is that the dynamic balances are the conductor’s, whose interpretation the engineers have presumably caught remarkably well.
And it is a superlatively characterful interpretation. Horenstein doesn’t just conduct this music, he caresses it into vibrant life – and to my mind he lavishes every bit as much attention on the accompaniments as he does on the leading lines. On the face of it, this may not seem a big deal. Yet, it seems to be something more than just, here and there, nudging a few notes up from the background; rather it’s a consequence of Horenstein’s “contouring” of accompanying lines. But the effect is significant, catching – and thoroughly enchanting – your ear; and, come to think of it, doesn’t it make a nice change to be surprised at hearing previously unnoticed details, not in a thoroughly modern stereophonic recording, but in a nigh-on 70-year-old monaural one?
The down-side, if I can so call it, is that after the first hearing these details have already settled comfortably into the musical landscape; to find them again I’d have to scrub my brain by some serious listening only to the recordings I know well. One instance that has stuck in my mind occurs in the third movement: Horenstein balances the components of Dvořák’s romping cross-rhythmic passages so shrewdly that the tapping toe is caught napping – something that’s far from always being the case.
The one surprise that I could have done without was a suspicion of unevenness in the slow movement’s sombre brass chords. Here, though, I am just picking a nit, and not even a particularly itchy one. Otherwise all is totally nit-free: a finely-considered, sublimely tender expression of Dvořák’s yearning music. The ensuing Scherzo fairly bounces along, impelled by some delectably clean articulation, the Trio section being relaxed in proportion. However, you may think that the scherzo section’s counter-subject, although cunningly moulded, is disproportionately slow; if so I’d refer you to my “point” made above, concerning the first movement’s tempi. Horenstein steers both middle movements – the one, so to speak, a “song” and the other a “dance” – with an appropriately light and deft hand.
Having had my “received wisdom” challenged, I was relatively unfazed by Horenstein’s umpteen eyebrow-raising tempo variations in the finale. Here, though, it’s more as though he’s “moulding” the music onto what he feels is the tempo most appropriate to the passage of the moment. Usually this sort of thing is a recipe for disaster (or, at least, for attracting accusations of playing fast and loose with the composer’s markings), but, although he does stretch the elastic a bit, at no time does Horenstein actually offend my delicate sensibilities. Without detriment to any of the customary physical excitement, Horenstein, building on the first movement’s “shadow of a question mark”, tends to bring out the music’s inherently mixed emotions. These climax quite properly in the coda, where two themes – the first movement’s first subject and the second’s main subject – collide in a grinding dissonance, which seems bereft of any mollifying garnish.
The Vienna Symphony Orchestra may be less than top-flight (shades of the old Suisse Romande, another fine, musically sensitive and endearing orchestra which many snobby folks rather unkindly looked down their noses at), but that is of little consequence: after all, Horenstein’s hosts of lovingly-turned touches would have gone for naught if the players hadn’t been right on the ball every inch of the way. Such evident, wholehearted commitment earns forgiveness for any number of so-called technical imperfections – not that in this instance the “any number” is at all excessive.
While I’m on that subject, there’s something that I simply can’t let pass without comment. One reviewer (famous but, regrettably, for far more than just the right reasons) complained, “The first movement introduction contains a near disaster, when the horns enter a beat early for all of their fortissimo eruptions beginning at bar 10.” The fact as stated is quite correct – but, if you compare Horenstein and A. N. Other, without referring to a score, although the difference is clear enough you’ll be hard-pressed to say who’s right and who’s wrong. Hence, to call it “a near disaster” is really to over-state the case by some margin. Moreover, it strikes me that someone clearly as attentive to detail as Horenstein would surely not have missed such a mistake made three times in quick succession. So, especially as there’s not even the slightest hint of it all actually falling apart, shouldn’t we admit the possibility that he might have committed this “near disaster” entirely deliberately?
This CD couples Dvořák with Janáček. Why isn’t such a pairing more common? After all, aren’t they arguably the champion Czech composers of successive centuries? I’m beginning to realise that Barbirolli’s 1960s concerts cost me a fortune: I recall his performance of Janáček’s succulent rhapsody, which gripped me from start to finish – and led to another urgent LP purchase. That was the now-legendary Ančerl recording (Supraphon SUA 50380, c/w Sinfonietta). Much later, tempted by the extra music (the Lachian Dances), I chanced my arm on a CD featuring the Slovak RSO conducted by Ondrej Lenárd (Naxos 8.550411). It turned out to be a bad bet.
Listening to Lenárd’s rendition of Taras Bulba in isolation, it seems a perfectly reasonable effort; I’ve no doubt that anyone living with this, as I did the Ančerl, would be quite content. But set Lenárd directly against Ančerl, and the former immediately feels distinctly prosaic and urbane where the latter burns with primitive, nervous passion – and, where appropriate, sounds as rich as a Christmas pud. Over many years, I’ve heard umpteen alternatives, but none (including Charles Mackerras’s effort on Decca/London 410138-2, which did impress me) has ever quite dislodged Ančerl from my affections. Are you expecting the punchline, “Until now”? Well, you’ll be disappointed. However, it’s very much to Horenstein’s credit that henceforward his Taras Bulba will find its way into my ears as often as Ančerl’s.
Commentators often contrast (with some justification) Taras Bulba and the Sinfonietta, suggesting that the late work sounds more like the work of a young man and vice versa. Yet, it’s not quite that clear a distinction: there’s a fair bit of the Sinfonietta that sounds “older” and plenty of Taras Bulba that sounds “younger”. In either work, the “younger” comes across in the more sparsely scored and rhythmically nervous music; the “older” emerges in the broader and more opulent passages.
The approaches of the two conductors are essentially complementary: comparatively speaking, Horenstein is the more “nervous” whilst Ančerl is the more “opulent” – and, odd though it may seem, if at any particular point some aspect seems lacking in the one, more often than not I will find it in the other! If only (I thought) these two had pooled their interpretative resources, what a performance that would have been! For instance, on the whole Horenstein more carefully cultivates his percussion colours. On several occasions he gives due prominence to thrilling contributions from the tympani that Ančerl leaves “embedded” in the texture. Ditto the bells, which give the music much of its “barbarian” character; early in the first movement’s quick section, they produce almost the clangour of an old-fashioned fire-engine alarm-bell, which, once you’ve heard it, you can also make out – if you listen very carefully – in the Ančerl.
Putting the boot on the other foot, Ančerl’s balancing of the organ, notably in the final movement’s concluding phase, is more astute, enriching the expansive peroration and making his huge final chords more massively imposing than Horenstein’s, in spite of the latter stretching them out to much greater length. Nor is Ančerl any slouch when it comes to realising Janáček’s white-hot, ear-piercing high string lines, only in this Horenstein does rather better than come in second. Then again, in matters of tempo, Horenstein is more radical whilst Ančerl tends to be more measured; yet neither falls short on the thrills front.
This Janáček recording is of a performance given at the 1961 Edinburgh Festival. The booklet doesn’t say whether it’s a BBC tape or an off-air capture, but (even allowing for the re-mastering) it sounds good enough to be the former. Audience noise, perhaps surprisingly, is at worst barely perceptible. The radio introduction is included on a separate track. Quite why I’m not sure; at a mere 41 seconds including applause it does no more than announce the work and the performers. The applause at the end is faded out before it’s even got going – might it have been more polite simply to have left it out?
Horenstein’s orchestra on this occasion was the Berlin Philharmonic, so the quality of the playing is not an issue – mind you, at Horenstein’s hair-raising tempo, even the BPO’s trombones have to be right on their toes in the first movement’s fast section. But the key point about the playing is that, perhaps surprisingly, it sounds remarkably idiomatic. Speaking purely musically, then, this disc is well worth the asking price, or maybe a fair bit more!
Now, what about Mr. Rose’s Magical Music Makeover Machine? Well, we all know that the basic idea of audio restoration is simple enough: to remove as much as possible of the stuff that shouldn’t be there (noise – clicks, hiss, hum, pops etc.) whilst losing as little as possible of the stuff that should be there (signal). Digital technology has provided tools that can deal with these, with varying degrees of effectiveness. One particular problem is that noise reduction software generally cannot distinguish ambience from “background noise” (tape hiss, gramophonic surface noise). In most cases a compromise must be made, between how much bathwater you throw out and how much baby you’re prepared to sacrifice; and it’s here that the delicate “art” of noise reduction comes in.
Thus far, the accent is firmly on preserving the original signal. If that’s regarded as one side of the coin, then the other side is the more controversial field of modifying the original signal to compensate for inherent shortcomings (some folk talk of “enhancing”, but to my mind this is not quite the same thing). To be fair, restoration engineers have always dipped their toes, however tentatively, in these waters, by tweaking their “tone controls” where, say, there’s an evident lack of treble, or adding a touch of reverberation if the noise removal has “dried out” the sound too much. This realm is where Andrew Rose’s “XR remastering” moves to centre-stage. Like noise reduction, XR is not a single process, but an ever-expanding suite of techniques, some of which Rose has developed himself and some of which are imported from third parties.
The foundation of XR is a technique for re-equalising a recording to correct for deficits in older recording and playback equipment. This is something that I (and, I’m sure, many other dabblers) have tried my hands at. My method was to take a recording’s sound spectrum, ignore the ripples of individual musical notes, draw a curve that evens out any unsightly bumps and hollows, and use that to calculate a filter to apply to the original recording. Occasionally, when I was very lucky, the result sounded “better”. Rose’s approach is similar, the crucial difference being that it’s minus all the guesswork. By comparing a fine-resolution spectrum (without smoothing) with a reference spectrum derived from several modern recordings of (ideally) the same music, he is able to obtain an exceedingly detailed and accurate filter profile. Is this a corruption of the original? To my mind, definitely not; if noise reduction, the removal of sounds that shouldn’t be there, is valid, then so is putting back sounds that should be there – provided only that you know what’s missing, which is where the reference spectrum comes in.
Apparently, particularly in the case of old “78s”, working with a copy of the original that hasn’t been noise-reduced it’s possible also to unearth valid HF harmonics buried within the noise. The recovery of these brings the immense benefit of substantially extending the old recording’s frequency range. Although may seem so, this is not a fiddle! – unlike synthesized harmonics (which extrapolate audible harmonic sequences), these were always there: they had been picked up by the microphones (or acoustic horns) and, because they were so weak, promptly buried within the noise. This is thus an extension of re-equalisation.
Using a digital resampling function to correct pitch inaccuracies, such as you get if the recording tape speed is slightly off, is straightforward enough, but what about oscillating pitch – the wow and flutter that are part and parcel of any analogue recording and playback? That’s a much taller order, because the pitch inaccuracy varies continuously, from sample to sample of the digitised audio. Hence I take my hat off to the German developers of the “Capstan” software that Rose has imported into his XR suite. I can think immediately of several of my old records that could do with this treatment. The “wobblies” are inevitably a bigger nuisance when your brain knows that they cannot possibly be interpreted as vibrato; hence, as Rose observes, this software can work wonders with old piano recordings.
Another problem is how to deal with recordings that were made in bone-dry acoustics, typically done deliberately in order to minimise the murkiness of the sound of old recordings when replayed on old record players. Such recordings, when remastered digitally and subjected to standard clean-up, sound, as you’d expect, “bone-dry”, like a picture whose subject is set against a jet-black backdrop. The standard fix is, of course, to add some synthetic reverb – and indeed, once upon a time certain, typically computer-based, playback facilities even offered a selection of pre-defined “stock” reverb options (some of the very names of which made a classical buff’s blood run cold! I invariably selected the “None” option). To be fair, applied with caution, good-quality reverb software, which uses the recorded signal as its basis and works by parametric modelling of “real” reverb, can produce results that are really quite good. However, Andrew Rose claims that the technique of “convolution reverberation”, which has been around almost as long, does a far, far better job.
Roughly (very roughly), this technique works in the same general way as ordinary reverb, only instead of a parametric model it uses a recording of the “impulse response” (IR) of an actual acoustic space. The IR is the sound of the reverberation caused by a single, sharp sound of wide frequency range (e.g. a pistol shot – presumably a blank). You can think of it as a “snapshot” containing all the information that defines the given venue’s acoustic properties, from which the actual reverberation of any sound can be extracted. The result of the process is a marriage of the original “dry” recording and the actual reverb of a chosen venue. If done artfully, the marriage would indeed be made in Heaven. Nevertheless and inescapably, this constitutes a case of adding something completely new to the original signal. Are we therefore to shun it as some sort of heresy? Well, that depends, doesn’t it?
We need to bear in mind that the use of an anechoic studio in the original was due to a choice forced, more or less, onto the recording engineers. If they had had the advantage of more modern equipment, then certainly they would have recorded in a live acoustic, wouldn’t they? Hence, the “validity” question boils down to how it sounds. If the result sounds wholly natural and completely convincing, then the process as applied is valid.
Finally (for now) there’s Rose’s “Ambient Stereo” component. This term is often misconstrued, possibly because the technique might be more accurately described as “stereo ambience”! Instead of adding anything new to or taking anything away from the original, this process merely redistributes some of it. It relies on a clever digital technique for extracting ambience information from a monaural original recording (other than, I suppose, a “bone dry” one!). This information is then, according to Mr. Rose, “spread across the stereo soundstage”, though how this is done I’m not sure. However, there is a big hint: Rose says that, if you switch your replay to mono (or remix your audio file to double-mono) the ambient stereo vanishes leaving the unchanged original signal. Unless I’m very much mistaken, this is possible only if the ambience added to the two channels is the same, apart from one channel being in opposite phase to the other, for only then will adding them together result in a big, fat zero added ambience across the board.
Again, is this a “valid” process? The purpose of Ambient Stereo is entirely laudable. I’m sure that pretty well anyone who has listened to mono recordings through headphones will have experienced the vague – or even severe – discomfort of hearing wonderful music hemmed in on both sides by a deathly blackness (or maybe “blankness”). If nothing else, taking out (or borrowing) the ambience of the central signal and “spreading” it either side does relieve the lateral gloom and, moreover, fills the blanks with something originating in the signal and still intimately related to it; I should think that, albeit only marginally, this will also clarify the musical signal. Here, though, there is no right or wrong answer: if it suits you then fine, if not you can switch it off; and you can’t say fairer than that, can you?
Apparently, there a lot more to XR; Rose says as much, but thus far hasn’t enlightened us, not even with a teasing little list. Nevertheless, what we do know about adds up to a very impressive armoury. As I see it, the only problem for us reviewers is that more often than not we won’t have the originals to hand for comparison purposes.
Only on the CD’s label do we find the modest declaration that the content is “ambient stereo”, though it seems clear enough that the recordings have been noise-reduced and re-equalised. Playing back the Dvořák, I listened carefully for signs of noise; only in quiet passages did I find any traces – just a merest hint of “crackliness”. I did compare the sound with an original, not of either of these particular recordings but of a contemporaneous Sibelius symphony (LSO/Anthony Collins). Comparatively, the latter’s sound quality was decidedly coarse. Not only is the XR sound smooth and even (seeming more like a recording of much more recent vintage), but the frequency range seems wider, feeling more “open” at the top and possessed of a notably firm bottom. At first the double-basses struck me as a trifle “boxy”, but that feeling rapidly faded as I started to marvel at their sheer substance, solidity and clarity; and that brought this thought, “Gosh, yes – in those days orchestras used to field eight double-basses, didn’t they?”
Other than it being in mono, this disc’s sound quality is very close to excellent – so much so that, in spite of Ančerl’s recording being in very good stereo, my choice of which Taras Bulba to pop into the CD player will depend only on which style of performance I’m in the mood for. The ambient stereo is certainly applied with discretion (it may not be of any great advantage to loudspeakers listeners, whose room acoustics will probably mask it). As often as not, I’m scarcely aware of it; but, if I am rash enough to switch to mono, I get a nasty jolt and hastily switch back. Rose is absolutely right in his claims for this technique: it may be a tad “artificial”, but in an almost self-effacing way it renders the headphones listening experience much more relaxing and pleasurable.
The 4-page booklet need never be removed from the jewel-case, as the inside is completely blank. Although the well-filled back says nothing about the music itself, this lack is admirably compensated by Misha Horenstein’s fascinating note about the conductor’s involvement with the works. Neither performance is a clear first choice for either work, but both are valid and thought-provoking alternative views, in vintage sound that’s been brushed up stunningly well – a thoroughly enjoyable disc that will hold its head up in any company.
Paul Serotsky
Previous review: John Quinn
