A Callas Recording Update

by Robert E. Seletsky

Since I evaluated the US pressings of EMI’s 1997-8 “Callas Edition” in “Callas at EMI: Remastering and Perception,” (The Opera Quarterly, Vol. 16, No. 3, pp. 240-255), which may be seen in an expanded form on this site, recorded Callas material continues to be remastered and reshuffled by EMI and other firms, with very mixed results.  The US pressings of the “Callas Edition” that I largely deplored have been discontinued, and only the superior Dutch pressings of that series are available.  At my suggestion in 1999, EMI corrected a disfiguring editing flaw at Callas’s entrance in the1997 remastering of the 1953 Tosca (EMI 56304).  However, the serious editing and pitch problems introduced into the 1997 remasterings of Rigoletto (EMI 66667) and six recital discs, to which I called EMI’s attention and which I cite in my 2000 article, remain unaddressed, despite EMI’s empty assurances.  Ironically, the corrected “Callas Edition” Tosca, though still in EMI’s catalogue, has largely been replaced by a 2002 remastering flying under two catalogue numbers in the mid-priced “Great Recordings of the Century” (GROTC) series (“Limited Edition”: 67756 [67759 “Angel” logo for US]; regular edition: 62890 [62893 “Angel”]), and a third in the budget “Historical Series” (85644)  The remastering, again by Allan Ramsay, does not have the 1997 error, and seems to be based on a different tape, but it begins at a pitch level of a1=435 Hz, drifting to standard a1=440 Hz by the end, thereby adding well over a minute to a 108-minute opera.  Pitch standards in opera houses during the 1950s were, if anything, higher, not lower, than a1=440 Hz, as original EMI Scala LPs attest.  Moreover, the 2002 remastering sounds vocally distant and artificially lush; further, sudden dynamic shifts at several edit points that were corrected in the 1997 edition are ignored in 2002.  One should avoid this multi-marketed muddle, and seek the now-corrected EU “Callas Edition” version, EMI 56304, if possible.

Encouraged by uninformed and Anglo-centric critical approbation of the GROTC Tosca, EMI, seeking to profit again from its greatest artist, initiated a third Callas go-round under the GROTC aegis. Recorded in Milan’s acoustically lifeless Cinema Metropole, the LP sound of the 1954 Norma wasn’t very vibrant; but whereas earlier EMI CD versions added a touch of “bloom” to the sound (47303 [47304 US catalog no.] and 56271), the 2003 GROTC version (62638 [62642 “Angel”]) removes all acoustic space, producing a harsh, compressed sound devoid of context, accentuating the overload distortion of the original. Moreover, a cache of pressings made from a defective master contains an unplayable final track on CD-3.  The 2004 GROTC issue of the 1953 Lucia (62747 [62764 “Angel”]) is likewise harsh and airless.  All of EMI’s 1953 Lucias on CD begin low in pitch, drifting upward through the opera.  Identical pitching, timings, and the retention of misguided CD pauses at some original LP side breaks—e.g., five frustrating seconds of silence between beats before “Ma dì, l’amato giovane” in Norma, Act I, sc. ii—indicate that, like the “Callas Edition,” GROTC versions may be remastered from 1980s DATs rather than original analogue tapes, if these even survive.  The GROTC Trovatore (62898 [62901 “Angel”]) has none of the sweetness, articulate bass, focus, or clarity of the LPs, and compared to previous CD incarnations, it is distorted at some points: the LPs had no distortion.  It is probably an indifferently modified version of the 1997 “Callas Edition” mastering which, particularly (if oddly) in its US incarnation, was the best CD compromise.  For EMI’s latest reissues of the 1954 Puccini (62794 [62795 “Angel”]) and Lyric/Coloratura (76843) recitals in the “Great Artists” series as well as the 1958 Verdi recital (57760) in the “Legend” series (EMI has endless “series”), we must contend with EMI’s confusing 1980s practice of filling out original LP contents with chronologically unrelated material.  Note that the Lyric/Coloratura disc is filled out with the 1955 arias from Vestale andMedea, making it identical to the first EMI CD release of the material in 1986 (47282): it even bears the same title, “Callas Sings Operatic Arias” (the first [Japanese] CD pressing used the word “Operatic” while later pressings shortened it to “Opera”).  My long-standing contention has been that EMI only remastered its Callas material digitally once, and that all successive remasterings are simply re-EQed versions of those original 1984-90 digitizations.  That belief is given further weight by the liner of this CD where we read: “This compilation ©1986” and “Digital remastering ©2005.”

The lapse of the 50-year EU copyright has legitimized “unofficial” LP transfers of studio recordings.  Callas LP transfers on Naxos appear on the heels of each copyright expiration.  So far, we have the 1953Lucia (8.110131-32), Puritani (8.110259-60), Tosca (8.110256-57), the 1954 Norma (8.110325-27), as well as the 1952 Cetra Gioconda (8.110302-04), and 1953 Traviata (8.110300-01).  Mark Obert-Thorn transfers the EMI sets, while Ward Marston handled the Cetra recordings.  Both men have impeccable reputations for splendid work with 78s.  LP transfers, however, present different challenges.

The ubiquitous use of the CEDAR de-clicking process, with its side-effect of “hearing” upper-octave transients as noise and subtly erasing them, is no problem with 78s, which seldom have much high-frequency content.  CEDAR’s unpredictable characteristic, however, is a distinct problem with full-range LPs.  Obert-Thorn’s LP transfers are scrupulously—even painstakingly—repitched and reworked to undo decades of EMI’s editorial mishandling; at last, the five-second dead spot in Norma is gone.  His transfers are untroubled by the shrillness and artificiality that often afflicts EMI’s Callas CDs, but his overzealous use of CEDAR and personal preferences result in the diminution of midrange clarity and a dulled upper range with a loss of transparency and articulation.  My own experimental, non-CEDARized, dubs of original 1950s Callas LPs to digital tape, using the same phono cartridge that Obert-Thorn employs (Stanton 500), produced results that retained the LPs’ characteristic sweetness, transparency, and focus, with no losses on top. Obert-Thorn’s mission to completely eliminate surface noise via aggressive interventionist processing impedes faithful LP reproduction, invariably making the “official” EMI CD versions clearly preferable to Naxos despite often serious reservations about EMI’s carelessness and sonic choices.

Upper frequencies in Ward Marston’s LP transfer of the 1952 Cetra Gioconda are likewise overly “rolled off,” losing much of the forward, crisp, transparent sound heard on the original Cetra LPs.  Still, hisGioconda is preferable to other available versions, all stemming from the artificially reverberant 1991 “SoundRebirth” Fonit Cetra remastering (Fonit Cetra CDO 8).  On CD, only the long out-of-print 1986 Fonit Cetra version, CDC 9, is accurate and pleasing.

The Naxos Gioconda set also includes the three 1949 arias from NormaPuritani, and Tristan, transferred by Marston directly from Cetra 78s.  They are very pure in sound, but again, I would have preferred a brighter, more articulate transfer even at the expense of retained surface noise (there is virtually none in Marston’s Naxos version).  In 1999, after Time-Warner acquired Fonit Cetra, the newly named Warner-Fonit issued the 23-minute “Maria Callas: Il primo disco,” (8573 80575) containing fresh transfers of the three 1949 arias; although not the quietest, these transfers retain the crispness and transparency of the 78s.  Not indicated by the annotator, “O rendetemi la speme…Qui la voce” is an alternate take from the one usually encountered.  Sadly, Warner-Fonit’s Gioconda and Traviata are clones of the same dreadful “SoundRebirth” Fonit Cetra masters heard on every other available CD aside from Marston’s Naxos transfers.  Further, in 2000, Warner inexplicably issued over-filtered, muddy versions of the 1949 arias in a disc that includes the standard Puritani take (8573 82241).  The gem of the Naxos transfers thus far is Marston’s Cetra Traviata: it is clearer, sweeter, more transparent and exciting than any other CD remastering of the recording.  One wishes all the Naxos transfers were on that level.

Threatened by EU copyright expiration, EMI, as well as other major labels, has gone to court in an effort to extend the copyright limit from 50 years to the 95 years prevalent in the US.  Concurrently, in order to remain competitive with Naxos, EMI issues parallel versions of its “official” Callas releases as inexpensive slim-line “two-fers” in its “Historical Series,” to which I referred earlier.  Thus far, the series includes Lucia(86197), Puritani (85647), Tosca (85644), Norma (86834), and even Cavalleria rusticana/Pagliacci (86830) in anticipation of Naxos’s forthcoming LP transfer; all are GROTC versions except Puritani andCav/Pag from the “Callas Edition,” in Spartan budget packaging.  Strangely, EMI markets these inexpensive versions not only in the EU but in the US where the Naxos releases are not sold because they would violate copyright laws.  I suppose, at the same prices, I would grudgingly say that the EMI issuances are somewhat better than Obert-Thorn’s better-edited but muffled, CEDARized Naxos products.  In the case of the 1953Tosca, had the EMI “two-fer” been a clone of the corrected “Callas Edition” version, I would have had no hesitation whatever in declaring its vast superiority; but, as the EMI budget version is copied from the off-pitch GROTC edition, neither of the two cheap editions is really acceptable.

Naxos publishes the original UK Columbia catalogue and LP matrix numbers in its booklets.  However, given the overall wretched quality of its EMI transfers in particular when compared to the splendid original LPs, as well as other sonic indicators, it seems unlikely that Naxos uses the actual pressings it cites; the information is evidently included only for historical purposes.  Briefly, LP transfers from the English firm Regis are amateurish, often made from uncorrected off-center copies that produce permanent “wow” in the CDs.  The advice in my 2000 TOQ article still holds: with few exceptions, Callas’s studio recordings are best experienced on vinyl; the older the pressing, the better.

After years of railing against “pirated” live Callas releases, EMI issued its own live 1958 Lisbon Traviata on LP as early as 1980 from privately owned tapes (EMI/HMV RLS-757, Angel ZBX-3910) , transferring it nicely to CD in 1987 (49187), though ruining it in a dreadful 1997 remastering (56330).  EMI’s live Callas efforts continued in 1990 with mediocre results effected by re-EQing random CD versions (Hunt/Arkadia, etc.—see my Spring 2000 TOQ article).  In 1999, Jackie Callas claimed sole proprietorship of her sister’s live performances.  Although the participation of many other artists in these performances made her claims ludicrous, she had lawyers send threatening letters to small companies issuing live Callas recordings.  Most ignored her; some were cowed.  Initially, she and a partner named Sakkaris issued CDs ineptly cloned from Hunt/Arkadia, Legendary, Rodolphe, Melodram, etc. on the Sakkaris/Diva label.  One may only surmise that the enterprise was terminated because of the discs’ poor quality.  In 2001, either out of charity or placation, EMI compensated Jackie Callas, now incorporated as “Marcal Investments” (MarCal = Maria Callas), to “license” anticipated live Callas recordings.

EMI/Marcal releases, as expected, are re-EQs, usually inferior, of pre-existing CD sources, their track placement and timings often identical to editions on Myto, Verona, Hunt/Arkadia, Melodram, Gala/Movieplay (a company intimidated by Jackie Callas), etc.  The 1953 Bernstein Medea (67909), from an out-of-print Verona CD set (27088/89), is muffled compared to its source.  Ricordi wouldn’t renew EMI’s license for the 1957 Ricordi/Mercury studio recording, one of EMI’s better-sounding “Callas Edition” efforts (66435—and only the EU pressing was impressive), which the live Bernstein set has replaced in the series.  BMG now owns the rights to the studio Medea but has not reissued it; interestingly, Decca has been reissuing recordings from the same “Mercury Living Presence” series.  The 1952 Covent Garden Norma(62668), either from Verona (27018/20) or AS-Disc (NSA 2801/3)—though probably the Verona set, is a few minutes too long because of noticeable downward speed drift; it is on three discs while consistently pitched versions on Melodram (26025) and Legato (LCD 130-2) are on two: Legato is the more correctly pitched and less filtered of the two.  The 1957 Scala Un ballo in maschera (67918) is drawn from a source with the first phrase missing, the second copied in its place, making the opera appear to open with two identical phrases in subdominant harmony; Golden Melodram’s version (GM 2.0009) has no such preposterous error.  EMI’s 1955 Sonnambula (67906), copied from Myto 890.06 (with identical timings) isn’t bad: EMI’s dulling of the highs imparts a certain sweetness.  The 1951 Mexico City Aida, copied from out-of-print Melodram 26015 (all the Spanish announcements are identically deleted/faded), is an improvement in one way: EMI consolidated the first two acts onto one 79’30” disc.  EMI also abbreviated some of Melodram’s infamous faked applause loops, though inadvertently retaining part of one that inaccurately overlaps a few notes in the final scene of Act IV.  EMI is often (though not always) more generous with track cues than its sources, and provides libretti with tri-lingual translations.

EMI also issued many of Callas’s recitals, some from awful CD sources, like the 1952 RAI recital released in 2002 (67922), which already appeared in EMI’s1997 2-CD “Live in Concert” (72030) from a far superior source.  EMI’s 27 February 1962 London concert (62684) is obviously cloned from Movieplay (100344-2), as the final cabaletta of the Anna Bolena scena is missing in both; the only other CD version, on Melodram 36513, from a more complete copy of the only original tape source, includes it (though Melodram bizarrely splices in the missing end of the Macbeth aria from the 1952 Scala performance).

Jackie Callas died on 3 March 2004, and EMI seems to have halted its live Callas releases.  However, Testament, a company that often licenses older EMI recordings for reissue, recently released the live June 1953 Covent Garden Aida conducted by John Barbirolli (SBT2-1355), claiming authority from the Barbarolli Society.  However, the off-air source is clearly identical to the one used for versions that have been available for a decade on the small labels Legato (LCD 187-2) and Golden Melodram (GM 2.0035); the engineer, perhaps tellingly, is EMI’s Paul Baily.  The Testament version, like EMI’s 1951 Mexico Aida, does have the benefit of integrating all of the first two acts onto the set’s first CD.

Cloned live Callas CDs are available for about $12 (US) per 2-CD set on Opera d’Oro.  That label’s 1955 Scala Andrea Chénier (OPD-1330), though limited by distortion in the original tape, seems to be the best-sounding version around; the 1955 Sonnambula (OPD-1139) and 1957 Ballo in maschera (OPD-1145) are identical to EMI’s; and the 1957 Anna Bolena (OPD-1285), cloned from the glorious GOP version (768-CD2)—a company sharing the same distributor, Allegro—should make EMI ashamed of its blurred, semi-tone-flat insult to this great performance (64941and 66471).  Opera d’Oro’s 1950 Parsifal(OPD-1236), 1951 Vespri siciliani (OPD-1291), 1954 Scala Vestale (OPD-1227) and 1956 Scala Barbiere (OPD-1264), and January 1964 Covent Garden Tosca (OPD-1142) are in mediocre sound, but are no worse than more expensive competitors.  The same observation applies to the Mexico 1950 Trovatore (OPD-1229), 1952 Lucia (OPD-1225), and Tosca (OPD-1248).  The 1952 Mexico Traviata (OPD-1226) and Rigoletto (OPD-1253), however, are far inferior to the brilliant LP masterings by BJR which included the original broadcast announcer and, in the case of Traviata, an inter-act interview with Callas in Italian; the long-discontinued Legendary Traviata CD seems to have been a transfer from a BJR LP, complete with interview.  The Opera d’Oro 1951 Aida (OPD-1250) must be avoided, as some of the orchestral music is unaccountably derived from unrelated performances.  The 1960 Poliuto (OPD-1228) is in considerably poorer sound than any other  version I know (EMI, Melodram, Verona) except the late, unlamented Virtuoso version.  The 1961 Scala Medea (OPD-1251) drifts a large semi-tone flat by Callas’s entrance, a problem in every edition on both LP and CD that no one has ever addressed.  Opera d’Oro’s Ifigenia in Tauride(OPD-1348) and Armida (OPD-1252) should never be preferred over Divina’s brilliant versions.  The 1955 RAI Norma (OPD-1301 and 7003), apparently owing to an error in the mid-1990s Fonit Cetra master (CDAR 2018, CDO 121) that acts as the source for all versions since, is missing thirty seconds in track 10 on the first disc.  Earlier masterings were complete, and Opera d’Oro plans to rectify the problem for both their budget and mid-priced versions of future pressings.  Needless to say, there has also been an endless procession of Callas clones from tiny companies with impossibly variable quality.

Several interesting and important items have come to light.  The first verse of “Casta diva” from a discarded take recorded for Cetra on 8 November 1949 has been issued on IDIS 6404; and, more significantly, the aria “Non si dà follia maggiore” from a live Il Turco in Italia performed in Rome on 19 October 1950, appears as a bonus track—and selling point—on Tima Club CLAMA CD48, a disc otherwise devoted to tenor Ulisse Lappas.  It is known that RAI recorded two broadcast performances of Turco (19 and 25 October 1950) and re-aired them months later, so there is some hope that this reappearance of one transcription disc may signal the survival of the rest.

The Lisbon Traviata (27 March 1958) has been mastered from a “new” source on Pearl GEMS 0228.  According to my information, the source is not new at all, but is a reworking of the oldest source: the original tape made by the Radiodifusão Portuguesa (RDP).  In the 1970s, it was copied to a chromium dioxide tape for conservation.  The RDP released a CD set (RDP 02/03), presumably made from this “safety” copy, in a very limited edition during 2000.  Unfortunately, either the transfer from the original tape to the chrome duplicate, or from the chrome tape to the CD (less likely), was carried out using the wrong EQ, producing severe sonic problems: greater dynamic range and clarity were offset by unlistenably harsh high frequencies and a very thin middle register.  Recently, a friend of the RDP offered engineer and opera scholar Mike Richter a copy of this “official” version with the expressed purpose of making it widely available at no cost.  Mr. Richter realized that the EQ was badly off and reworked the CDs until they sounded better, a task he performed gratis.  He circulates his corrected version free of charge, as stipulated by the donor, directly from his web site, http://www.mrichter.com/ , in CD-ROM format; the only fee is the nominal price of the CD-ROMs themselves.  Both individuals believe strongly in preserving important musical documents while keeping them from profiteers.

One recipient of the Richter-EQed version, however, passed a copy along to a Pearl engineer and Pearl issued it commercially, apparently doing no more than adding a few track divisions—and without crediting Mr. Richter (which he doesn’t seem to mind).  If true (and as the information was made public by Mike Richter himself, there really is no doubt), the fact that Pearl is selling this item—at a high price—negates the terms and intent of its original availability: so much for keeping it from profiteers.  Moreover, while Mr. Richter’s re-EQ tamed the abnormally harsh upper register, the “sucked-out” middle is still very problematic. Callas’s voice is often nearly unrecognizable, the recording suppressing the richness of her middle register, which even in her darkest days, never deserted her.  The problem is completely clarified in a direct comparison with EMI’s more natural, balanced 1980 LP (HMV RLS-757, Angel ZBX-3910) and 1987 CD (49187) versions—though not EMI’s horrible, ruined 1997 “Callas Edition” remastering.  EMI’s source was a tape copied directly from the original for tenor Alfredo Kraus right after the performance, without any EQ errors.  Evidently, Kraus first gave it to record producer Ed Rosen, who issued it on LP in the 1970s and later sold it to EMI.  Although the Kraus source has the reduced upper-frequency clarity and dynamic range typical of old analogue tape copies, it is preferable.  Sadly, the 1987 EMI CD set is out of print and hard to find; one should still seek it out if possible, but if one had to choose between the currently available options, I suppose the RDP-Richter(/Pearl) version would be preferable to EMI’s 1997 disaster.

The work of Divina Records is consistently without peer, the finest presentation of Callas material on CD. Divina has discovered new sources and performances, remastered them with a level of care unknown in the CD era, and produced results of overwhelming beauty and excellence, including a 1952 Armida (DVN-16) finally in listenable sound and at correct pitch, with over twelve newly-discovered or rediscovered minutes of music; and a 1955 Scala Norma (DVN-17) from a superb first-generation tape (though the first phrase of the overture is pitched slightly off, to be corrected, I am assured, in the next pressing run). Moreover, Divina has unearthed and issued, in impeccable masterings, rare materials that were previously unknown or had existed only in marginal or fragmentary condition; examples include the most complete version of the 1953 Trieste Norma (DVN-3), the Teatro Colòn Recordings from the original paper tapes (DVN-12), and two volumes of Unknown Recordings (DVN-1 and 4).  Most recordings are issued as enhanced CDs that reveal articles, high-quality photos, and even video clips from extant Callas performances when played on a computer.  Divina is the BJR of CDs.

Robert E. Seletsky is an independent scholar and former baroque violinist and music director with published work appearing in New Grove II (2000), Early MusicThe Opera QuarterlyRecent Researches in the Music of the Classical EraThe New Harvard Dictionary of Music, and elsewhere. E-mail: memres@sprintmail.com

This article is heavily revised and expanded from The Opera Quarterly, Vol. 21, No. 2 (Spring 2005), pp. 387-391, and appears through permission of Oxford University Press.

©2005 by Robert E. Seletsky
©2005 by Oxford University Press