Support us financially by purchasing this from Support us financially by purchasing this from

Richard STRAUSS (1864-1949)
Salome, Op. 54
Birgit Nilsson (Salome); Eberhard Wächter (Jochanaan); Gerhard Stolze (Herod); Grace Hoffman (Herodias); Waldemar Kmentt (Narraboth); Josephine Veasey (Herodias’ Page); Liselotte Maikl (A Slave); Paul Kuen (First Jew); Stefan Schwer (Second Jew); Kurt Equiluz (Third Jew); Aron Gestner (Fourth Jew); Max Proebstl (Fifth Jew); Tom Krause (First Nazarene); Nigel Douglas (Second Nazarene); Zenon Kosnowski (First Soldier); Heinz Holecek (Second Soldier); Theodor Kirschbichler (A Cappadocian)
Wiener Philharmoniker / Sir Georg Solti
rec. 1961, Sofiensaal, Vienna
BD-A in Dolby True HD LPCM 2.0
Hardcover booklet in English, French & German, including synopsis & libretto
DECCA 483 1498 CD/BD-A [100:36]

Elektra, Op. 58
Birgit Nilsson (Elektra); Regina Resnik (Klytämnestra); Marie Collier (Chrysothemis); Gerhard Stolze (Aegisth); Tom Krause (Orest); Tugomir Franc (Orest’s Tutor); Margareta Sjöstedt (Confidante); Margarita Lilowa (Train-bearer); Gerhard Unger (Young Servant); Leo Heppe (Old Servant); Pauline Tinsley (Overseer); Helen Watts (First Maid); Maureen Lehane (Second Maid); Yvonne Minton (Third Maid); Jane Cook (Fourth Maid); Felicia Weathers (Fifth Maid)
Wiener Philharmoniker / Sir Georg Solti
rec. 1966, Sofiensaal, Vienna
BD-A in Dolby True HD LPCM 2.0
Hardcover booklet in English, French & German, including synopsis & libretto
DECCA 483 1494 CD/BD-A [107.50]

There’s an old adage in law and politics that you should never ask a question you don't already know the answer to. This has wider reach, as I found to my embarrassment when co-convening a seminar on music and sound recording in the late 1970s. The principal guest was John Culshaw, producer of these recordings of Salome and Elektra. I was managing the technical side of the discussion, and decided on a whim to ask John to explain Decca’s ‘SonicStage’, which is touted for both productions. Expecting some elaboration about singers moving about on numbered squares to achieve a kind of virtual reality, John simply looked bemused and replied “well, it doesn’t mean anything, really” and adding, in effect, that it was just an advertising slogan [1].

Such are the myths and misconceptions of sound recording and reproduction. Add the dopamine buzz of music, and a strange chemistry happens which makes listeners want to believe a whole lot of things they rationally shouldn’t. It soon sank in what John was talking about, as by then having accumulated some experience in sound engineering myself, I realised that especially in the early days of stereo, what Decca were doing was really the only practical way of achieving natural perspectives, and indeed its competitors were doing much the same thing. More on that later.

The enduring success of both these recordings is not surprising when you consider their makeup. Within the charged atmosphere of Strauss’s operatic realisations of Wilde’s play for Salome and Hofmannthal’s libretto for Elektra, there is a synergy between conductor, orchestra, soloists and production team that delivers something uniquely special. Georg Solti was very much in his element, especially in Salome where his natural instincts for precision, relentless tension and irresistible drive pull the listener almost bodily into the dystopian aura of Herod’s court, never letting go until the anti-heroine is crushed beneath the soldiers’ shields. For that part, Birgit Nilsson is sui generis; whether or not one feels in the aural presence of a seductive teenager, the glorious singing and sonic fantasy of it all create their own spectacle. Her voice’s typically hard edge only enhances Salome’s wickedness, and its maturity and power leave you in no doubt her depraved temptress means business. Nilsson’s final scene is a tour de force, one of the greatest in recorded opera, where you cannot help but be in Salome’s thrall; she is as eerily alluring as she is repulsive, and you are a willing voyeur to this scene of morbid horror, a piece of sensual grand guignol as John Culshaw described it. At its culmination where Salome kisses the lifeless Jochanaan’s lips, Nilsson is spotlighted to be almost whispering in your ear – or are you now Jochanaan? – an effect some might say is a touch of genius, others perhaps cheap and tacky, but it’s there, and a reminder of one of the most creative periods in gramophone history for the recording of opera, when producers such as Culshaw believed that special measures were needed to bring the full impact and imagery of opera into the home.

With much the same venomous atmosphere and of similar scale, Solti’s Elektra is also a vivid experience. Likewise, a hard-edged Nilsson as the eponymous femme fatale brings out all the dominant brutality of the role. But while there is nothing conciliatory in Salome, the later work is more emotionally layered, and here Solti and his principal star are rather caught out when the venom recedes for the Recognition Scene between Elektra and Orestes. In softening her voice, Nilsson goes a little flat while Solti, gung-ho elsewhere in pouring on the orchestral bile, fails to fully convey the languorous warmth of this moment, suggesting instead thwarted adrenaline. And speaking of the orchestra, the Vienna Philharmonic play superbly throughout, with the cohesion of a single organism.

If it appears I’m neglecting the other elements, my purpose here is to highlight the core common features that define these recordings. Suffice it to say that the casts for both could hardly be bettered; witness, for example, singers of the calibre of Helen Watts and Yvonne Minton in minor roles. What’s possibly more important is how well suited these singers were to the production approach, in amplitude and power throughout their ranges, to achieve natural balances and sustain dramatic flow. In this they succeed handsomely, and were it not for some minor hiccups in Elektra, it would be difficult to separate the two productions. As it is, I believe Salome to be the greater achievement.

Both operas were recorded in Vienna while Decca was also engaged in the Wagner Ring project. Between Salome and Elektra, the same team recorded Götterdämmerung, using a refurbished control room in the Sofiensaal, so there were differences at least in the technology supporting each production, if not in the production approach. A very good behind-the-scenes glimpse is given in Humphrey Burton’s The Golden Ring, made during the recording of Götterdämmerung. The frequent criticism of recessed voices in these recordings, which I admit doesn’t bother me, did apparently trouble the singers, as Burton observes during a control room playback scene that they “are more concerned with being heard through the orchestra”. As the vision shows, the singers are on a stage (the ‘sonic stage’, if you like!) behind the orchestra, covered by three spaced microphones at the front, and being guided by assistant producers to pre-determined positions to deliver their lines. As I understand it, Decca preferred going straight into stereo even when multi-tracking was available (a multi-track machine is shown in the Sofiensaal footage, presumably for back-up [2]), rather than adjusting balances post-production. What you essentially hear, therefore, is two overlaid stereo images, one for the voices and one for the orchestra, balanced and preserved in real time for all time [3]. It strikes me that the singers in Elektra are a little more forward in the mix, but that may be due to the sparer orchestral textures – the sound in general is not as sumptuous as it is for Salome. Certainly the ‘special effects’, such as the over-dubbed voices of anguish during Orest is tot!, leap out rather alarmingly and unnaturally.

Decca were also renowned for driving their tape technology hard, and the generally high levels with frequent hints of tape saturation suggest these are highly modulated recordings. The natural compression kept up a high level of intensity, a forerunner perhaps of today’s ‘loudness wars’. The sustained impact worked well in LP days, keeping surface noise relatively subdued, with the occasional congestion nothing unexpected as end-of-side distortion was the norm anyway. Digital technology ruthlessly exposes such imperfections, but for the most part, the recordings hold up very well.

As best I can tell, this is the third time these recordings have appeared in digital format, following previous reissues of Salome in 1985 and 2006, and Elektra in 1986 and 2007. For new buyers or collectors who do not already have these performances on their shelves, this is not a problem – if you have any regard at all for these Strauss operas, these recordings couldn’t be more strongly recommended, not only on musical and production grounds, but as significant pieces of gramophone history.

For those who already have the previous releases, however, there may be perplexing questions. The 2006/7 reissues were in the Decca Originals series, promoted as high resolution (hi-res) digital transfers of the original analogue mastertapes which, by implication, the 1985/6 transfers were not. How so? Leaving resolution aside for the moment, I recently came across a 2015 Guardian article on the CD’s history, in which an industry executive is reported as saying: "We’re making a huge mistake. We’re putting studio-quality masters into the hands of people."  I'd suggest in fact this penny dropped industry-wide from the outset, as indications are that copy-masters were used instead of originals for many early reissues [4]. The majors would have been awash with these copy-masters, having been produced for making LPs, pre-recorded cassettes and the like, and were ready-made for transfer to CD. Copy-masters, though, can have significantly greater noise, distortion and compression than the originals, sounding more strident, muddled and opaque. Is that the case with the 1985/86 transfers?

Comparing different transfers of the same recording with any rigour requires, at least, that it be done blind, and at identical volume levels for each sample. Too often I read about listeners, including reviewers, doing comparisons in such a casual way that preconceptions and existing beliefs will overwhelm good judgement, not to mention group-think in collective situations. Cutting to the chase, I compared samples of my 1985/86 transfers with the current CD transfers (I’ll get to the BD-A later). Blind comparison [5] on loudspeakers showed the earlier transfers to be at a slightly lower level, but once equalised to the same volume, I couldn’t reliably distinguish them from the new transfers. Only on headphones could I detect some consistent differences, relating mainly to image instability on the earlier transfers which I would attribute to some tape dropout – if indeed the 1985/86 transfers were from copy-masters, they were very good ones. Omnipresent was the trademark Decca stridency of that era, which I’ve commented on elsewhere. Here, at least, it adds to the shrill and sinister atmosphere.

Now, you might be saying, surely the hi-res remastering would make a difference? Well, not necessarily – as a CD at least, you’re still hearing it in standard format. And further, considering just the transfer of stereo analogue tapes from this era to digital media, the CD is prima facie more than adequate, comfortably exceeding both the frequency and dynamic ranges of the older technology [6], and effectively transcribing without error. This assertion can be extended to all stereo sources, as indeed it has been vigorously argued [7] in the online trenches, and largely supported by the research studies [8] about the limits of human hearing and what listeners can and can’t differentiate. Nothing defines the quality of what you hear better than the quality of the original source material, but ‘remastered’ [9] is such a magic word for some, anticipating all sorts of miracles that, frankly, won’t happen, but you can’t stop people hearing what they want to hear. In this case, what more would you want than an exact transcription of the original Decca tapes as heard during playback in the Sofiensaal control room? No tricks, really, just use the actual original masters, in carefully managed digital transfers [10]. Personally, by the way, I don’t mind tape-hiss.

Which leads to the BD-A disc included with both sets. With its hi-res content, BD-A contains a lot more information than CD [11], and is technically a more ‘perfect’ copy of the analogue source, but can you hear it? The research mentioned above suggests the audible difference of any hi-res source is marginal at best, so the largely anecdotal claims about BD-A transfers, such as ‘revelatory’, seem rather extreme. If it were true for these releases, it would be fair to ask whether different sources or signal processing had been used. Comparatively, the main functional advantage of BD-A in this case, being of a stereo source [12], is that the whole of each one-act opera can be accommodated on its BD-A disc, and heard without a break. Artistically and emotionally, this is a considerable advantage, although in the case of the Elektra BD-A, there is a pause (effectively between the first and second CDs) which deviates from Strauss’s score. For those however without a score in hand, the pause, lasting barely more than a second, would hardly register were it not for the sudden loss of ambience following Elektra’s Worüber freut sich das Weib?, momentarily creating a kind of a sonic vacuum.

Blind comparison of BD-A versus CD sources is a physically fraught matter unless you’re organised for it, including tolerant, helping hands. An indirect [13] but less onerous test of whether BD-A sound differs from CD is to run the Blu-ray player (or DAC) outputs alternatively through a ‘CD quality’ A/D/A chain, carefully equalise the levels, and have someone switch between them, noting the listener’s preferences. This I did [14], sampling both PCM and Dolby True HD layers, and could only differentiate between the two signals at high levels on very quiet passages by the slightly higher noise in the CD simulation, which had no added dither. No revelations, I’m afraid, which is not to say definitively there are no audible differences between the BD-A and CD versions of these reissues, but my general listening contained no ‘Eureka’ moments of added enlightenment. Indeed, if there were any, my first hunch would be that there were differences in the mastering of the two versions, including interventions to subliminally suggest a difference – some subtle EQ, for example.

Reviewing recorded music is by and large a subjective business. On both performance and sound, it’s more about what is liked than whether any absolutes are satisfied. When it comes, however, to comparing transcriptions of the same original recording, stricter methods and criteria can be applied which are meaningful both in the objective assessment of the result, and the validation of any claims made for it. Or as Vernon Handley said about his side of the music game: “you shouldn't fraudulently convince people that they have heard what they haven't” (Sunday Times, 1984). I can understand why the recording companies manage these classic archives as they do, for example to keep copyright control and to maintain a reasonable income stream from them, but I do wish they would be a little more upfront about what they are actually offering for sale. Too often, it appears, we’re getting the ‘same old, same old’ material, possibly a fresher generation of the original master, cynically dressed up as a technological advance.  How would we actually know when we've struck gold [15]? As with the imaginary ‘SonicStage’ [16], we now have Blu-ray ‘Pure Audio’ dangled evocatively before us. Pure audio? Well, I suppose there are no pictures – moving ones, anyway. What next for Salome and Elektra? Why, vinyl of course! Nothing actually, objectively, superior there, but riding a wave of public sentiment never hurt any entrepreneurial spirit, or to quote another old adage: follow the money.

If I appear to have strayed from my mission here, it’s only to underline that recordings such as these are so important that everything about them now deserves to be heard. Perhaps it is, but my experience in the audio game and as a record collector has bred both knowledge and scepticism which look critically at all such reissues and ask ‘is this it?’. The signs that we are at last hearing the studio originals can be confusing, particularly with highly modulated Decca tapes such as these. But I’m reassured at least by the rock-steady, well-defined stereo images of these new releases that they’re about as close as we are likely to get. In answer to my earlier hypothetical to collectors who already have the previous issues, as well as a Blu-ray player, the advantage of playing each opera without a break is considerable, and for whatever other virtues you perceive for the BD-A medium. The new sets are also attractively packaged and should take up less shelf space. In short, buy them.

Des Hutchinson

Previous reviews: John Quinn (Salome – Recording of the Month) ~ Paul Corfield Godfrey (Elektra)

Notes 
[1] In his later memoir Putting the Record Straight (Secker & Warburg, 1981), John Culshaw relates that this ‘spoof’ originated when he suggested, dead-pan, that in Decca’s Rheingold recording, the voices of the Rhinemaidens could be heard coming from below the Rainbow Bridge. This was swallowed by most of the music press and, now emboldened, Culshaw’s team suggested for Salome that they had invented an entirely new approach to operatic recording, calling it ‘SonicStage’. Decca’s advertising machine embraced the slogan, and it became unstoppable. But as Culshaw reflected “there was not a jot of difference between Salome and any other opera we had recorded in Vienna for the past three years”. While he somewhat regretted the episode, he saw it as “a kind of private revenge on those critics who persisted in parading their prejudices”.

[2] According to Mike Gray in The Decca Sound: Secrets of the Engineers: “Decca always wanted to mix to two tracks, although they had four-track backups as early as the late ’50s”. This begs the questions: What happened to these four-track and other multi-track masters of this and the later analogue era? How many are still available for re-mixing and re-mastering, or were the majority destroyed or re-used once the original stereo master had been produced?

[3] Once multi-tracking became the industry norm, opera recordings usually involved one microphone per singer, with its own track, and then mixed with pan-and-fade to the stereo master in post-production. If the multi-track tapes were retained, these could potentially be re-mixed to a new, and superior, digital stereo master, since the original analogue stereo master was effectively a tape copy.

[4] My Exhibit A for this practice is the first CD reissue of Otto Klemperer’s Beethoven symphonies on EMI. As an experienced recordist, I suspected from the flat, lifeless and steely sound that these were not from original masters, but the real tell-tale signs that these were from the copy-masters used for cutting the LPs (SLS 788/9) were the rapidly truncated reverberation at the end of each movement, and other artefacts such as switching ‘clicks’ in identical places on both the LPs and CDs. Later re-issues of these recordings by EMI (e.g. in the Klemperer Legacy and GROC series) had none of these issues, and sounded noticeably ‘fresher’.

[5] Tracks used for comparison: Salome CD1: (1) & (11), CD2: (3) & (4); Elektra CD1: (1) & (8), CD2: (1) & (8). My selection was limited by differences in the cueing of several tracks between the 1985/86 reissues and the new ones, which eliminated those tracks from blind comparison. The earlier sets were cued with greater precision. I had wanted to use the same Salome track referred to in the September Listening Studio report (Ah! Du wolltest mich nicht deinen Mund), but the new transfer begins with a fragment from the previous track. I also noted this track has frequent tape overload which is manifestly audible on all digital media.

[6] For example, the Studer A62 professional stereo tape recorder, introduced in 1965 and a common standard at the time, specified a frequency range of 15kHz and a signal-to-noise ratio of 55dB at 15 inches-per-second. Fully modulated, this gave an effective dynamic range of about 65dB. This is comfortably covered by the 16-bit, 44.1kHz sampling rate of the CD, which provides for a dynamic range (effectively a dynamic window) of 96dB and a frequency range of 22.05kHz. With dither (see first article referenced in note [7]), this can be further improved.

[7] For example, a widely cited 2012 article on 24/192 downloads, and “why they make no sense”, by Christopher Montgomery, a digital audio engineer who heads the non-profit Xiph.org Foundation that’s responsible for the Opus, Ogg Vorbis, and FLAC digital audio codecs. As with any article of this kind, self-interest must be taken into account, but I find his arguments to be based on solid science and research evidence, and generally free of hyperbole. Note particularly his coverage of anti-aliasing filters and intermodulation effects. Readers however may prefer to side, say, with the 2010 paper on hi-res audibility by Dr. Hans R.E. van Maanen.

[8] Among a plethora of studies on hi-res audibility, the Audio Engineering Society (AES) has in recent years published two widely discussed papers: the 2007 study by the Boston Audio Society (BAS) entitled “Audibility of a CD-Standard A/D/A Loop Inserted into High-Resolution Audio Playback”, and the 2016 study by the Queen Mary University of London (QMUL) entitled “A Meta-Analysis of High Resolution Audio Perceptual Evaluation”. The BAS study, involving actual listening tests, found that subjects effectively couldn’t differentiate between hi-res (SACD) and CD quality, while the QMUL study, a meta-analysis of several smaller, less conclusive studies, found that subjects who’d received “extensive training” were able to differentiate better than those who hadn’t. The overall difference in statistical significance reported by the studies is not large, the BAS result of 50% implying subjects were simply guessing, while the QMUL result of 51% (untrained) and 62% (trained) suggests subjects were still guessing much of the time. The QMUL study devotes some effort to critically scrutinising the BAS study, noting that its results had been disputed, including through the online forum www.hydrogenaud.io. But equally, the QMUL study has been disputed on the same forum. The AES Journal has plans for a special issue on hi-res audio, with a target publication date of July/August 2018. As they say, watch this space.

[9] Not to be confused with ‘restored’, ‘re-imagined’ or other verbal sophistry the recording industry is prone to use. I simply refer to a digital transfer that gives the home listener, in this case, an audibly accurate, warts-and-all reproduction of the original analogue stereo masters, or a replication of the original stereo mix if original multi-track masters are still available.

[10] For one brief, shining moment this may have happened with the earliest CD reissues of Bruno Walter’s stereo recordings with the Columbia Symphony Orchestra, allegedly digitised directly from the original mastertapes. I don’t believe they’ve ever sounded so good. This maybe caused the industry to take fright, and thence exercise ‘quality control’ on its own terms. Their saving grace may have been the relatively low demand for these particular reissues, and that the original recordings were good but not great.

[11] The 24-bit, 96kHz sampling rate cited for these transfers gives a frequency range of 48kHz and dynamic range of 144dB which, as observed in note [6], would greatly exceed the requirements for covering the frequency and dynamic ranges of the original tapes. Arguably, therefore, a large portion of digital data on the BD-A is of no musical or sonic value. There appears to be a misconception of associating audio resolution with video resolution, i.e. ‘pixel density’, when in fact it is quite different. In particular, the expansion above 16 bits was primarily to give engineers more headroom for digital recording (Decca, I understand, were using 18 bits from the outset). In a fully modulated 24-bit recording, only the first 14 to 16 bits are likely to be musical signal, with the rest occupied by sub-audible noise (the 17th bit is at -102dB). Shaving off the lowest 8 bits and perhaps adding dither should therefore make no difference for normal listening. Those wishing to challenge this are advised to first take out hearing insurance.

[12] I am excluding any ‘special features’ of coding formats or players, such as the fabrication of subwoofer information, from my strict stereo-vs-stereo comparison for both types of media, to be consistent with the configuration of the original source material.

[13] This is the method used for the BAS study (see note [8]).

[14] Sadly, I can only offer a sample of one, namely myself. Why doesn’t the assistant also do the listening test? Usual response is “Who cares?”! I’ll also add that those enquiring after my listening equipment will be disappointed. To quote the old sales response about the horsepower of a Rolls-Royce, it’s “adequate”. As I’ve also designed and built much of it myself, indulging in ‘brand wars’ will be unproductive. There’s also futility in me saying I can’t hear any revelations on System X, when you will say “Oh, but you can on System Y” - perchance I then get to hear System Y and, rigorously tested, I still record a null result, you’ll then likely say “Oh, but you need a hearing test”. And so it goes on and on. I’m open to discussion, though, on the related science, technology and research methodology issues.

[15] Without knowing what I’m actually getting in terms of originality, my preference is always to purchase the SACD or BD-A version of a reissued recording, if available. Two principal reasons: access to multichannel sound (if so encoded), and potentially better production values. Especially for older analogue recordings in multichannel, the companies are compelled to go back to original studio masters instead of just digitising another n^th generation stereo copy for the next CD release, and calling it ‘remastered’. The BAS study (see [8]) noted that poor production values had contributed to the CD's bad reputation and, conversely, that hi-res equivalents are “made with great care and manifest affection ... that sound as good as they can make them”. The authors further conclude, as a result of their findings, that “all of these recordings could be released on conventional CDs with no audible difference.”

[16] Curiously, the new booklet for Salome contains an abridgement of John Culshaw’s essay on ‘SonicStage’ from the original LP set, apparently oblivious to what he later said about it (see note [1]).  (No mention of ‘SonicStage’ is made in the 1985 CD reissue.)  This, together with the rather imprecise cueing on both new sets and the unscripted gap in the Elektra BD-A, suggest some lack of care and artistic oversight.