Callas Recording Recommendations 2019 (with revisions 4/2024)

—Robert E. Seletsky

I have frequently been asked to revise my list of recommended Callas CD transfers. Although my outdated 2005 list on the Divina Records website simply provided titles and catalog numbers, the matter has become more involved for me, reflected here.

Live Callas

Compared with 2005, there are fewer companies issuing live Callas, but most offerings are better than the often-dreadful earlier options. On the other hand, not everything has reappeared. The principal labels are [1] Divina, abbreviated here as DVN, the most scholarly and perfectly produced issuances ever to grace live Callas material. Lest it seems like I am advertising for Divina, I have worked with Pablo Berruti periodically on matters of pitch and sound for over twenty years (I am a professional musician and musicologist). I still marvel at his brilliant researches, exciting discoveries, technical wizardry and command of early technologies (e.g., paper tapes, 16-inch acetate transcription discs) on which he must often play the many original sources he has amassed, as well, of course, as his love for Callas. [2] After Divina, there is Ars Vocalis who occasionally sells his discs on eBay, referenced here as AV: transfers from the best LP and other sources by an encyclopedically informed Callas scholar, some of whose written work appears on the Divina Records site, and who was largely responsible for the enormous amounts of supporting materials accompanying many of DVN’s releases. I have known AV for decades as well and have collaborated on such matters as sources, pitch, sound, etc. As a rule, AV will not overlap with DVN. He may be contacted at (note that ‘info.’ is important in the address as ‘ars.vocalis’ is someone else’s g-mail address). [3] Myto Historical Series from ca. 2006 onward. Myto can be careless, but they have issued some things in good form, cloned from fine earlier releases drawn from BJR LPs, and have issued a few things in pitch-corrected versions for the first time. Incidentally, the best prices in the US for Myto are currently offered by Norbeck, Peters, & Ford: [4] Warner Classics, the arm of the multi-national corporation that acquired Cetra and EMI, has issued a number of live Callas performances at breathtakingly low prices, all re-equalized previous releases (despite their hype), generally EMI which were already re-EQs of others’ CDs. ‘Art & Son’, the firm that did the re-EQs, seems to have treated all issuances the same: they rolled off the top and boosted the upper bass. Some are passable for those on a budget, but the sound is often far from optimal; and for beginners, the lack of libretti is problematic in heavily edited and oddly translated performances. There are a few other labels, like Testament, etc., issuing good discs.

DVN owns all the original Mexico material so other choices for those performances listed here are only stopgap measures until Pablo Berruti issues them, as he has done with his amazing 1952 Puritani. Of course, he owns all the BJR tapes as well, and thus far, has issued a number of important historical documents from them: e.g., Anna Bolena, Berlin 1955 Lucia, Cologne and first Edinburgh Sonnambulas, Ifigenia in Tauride, RAI Recitals.

Not all CDs that I recommend are in print: e.g., Gala, Melodram, Eklipse. DVN and AV are burned-to-order CDs or downloads, while Myto, Warner, Opera d’Oro, Gala, EMI, etc. are pressed CDs.

Bold indicates preference. Underline indicates current availability if the first choice is out of print (OOP). Dates are given in US style: month/day/year or month/year.

Complete Operas

Norma, Buenos Aires 6/17/49 extant excerpts: DVN-12.
Norma, Buenos Aires 7/9/49 extant excerpts: DVN-12.
Norma, Mexico City 5/23/50: AV (same as OOP Urania URN 22.133); Myto 00262.
Norma, Covent Garden, 11/18/52: AV, the only version in accurate sound and truly complete, from the original tape in the Omega Opera Archive; contains bonus of discarded 1949 Cetra ‘Casta diva’ verse in somewhat clearer sound than IDIS 6404 [‘Rarities by Maria Callas’].
Norma, Trieste 11/19/53: DVN-3, a brilliant effort of research. Earlier, now-OOP versions from other companies were composites containing material from other performances.
Norma, Rome RAI 6/29/55: AV for original dry sound; Myto 00140 has a hint of reverb.
Norma, Scala 12/7/55: DVN-17: the only version from the actual, single original tape.
Norma, ‘Rome Walkout’ 1/2/58: DVN-7, currently unavailable; Myto 00151.
Norma, ‘Sediziose voci’ excerpt, Paris 5/64 rehearsal: DVN-4 enhanced with video file; also in EMI DVD ‘The Callas Conversations, Vol 2’.
Norma, Paris 5/17 & 21 & 29/65, extant excerpts: DVN-10 &15 ‘The Last Normas’.
Pirata, New York 1/27/59: DVN-21, the only version truly from the original tape. Warner is a pitch correction of EMI. NB: keep Italian-English EMI libretto reflecting the strange and brutal cuts.
Puritani, Mexico 5/29/52: DVN-23, cause for rejoicing after muddy, cut, conflated Melodram 26027.
Sonnambula, Scala 3/5/55: Myto 80006. EMI 67906 is filtered Myto. Warner’s EMI re-EQ is even more muffled as well as distorted, thin, and strident on top.
Sonnambula, Cologne 7/4/57: DVN-22, finally with all proper materials and no spurious ones.
Sonnambula, Cologne 7/6/57: DVN-8, same comment. New evidence that July 4 & 6 dates are reversed.
Sonnambula, Edinburgh 8/21/57: DVN-9 from BJR tape; Testament SBT2-1417, from Walter Legge’s tape remastered by Paul Baily, whose editing is not to be trusted; OOP Melodram 26037 is filtered, dull.
Sonnambula, Edinburgh 8/26/57: DVN-6.

Medea, Florence 5/7/53: AV from Fonit Cetra LPs; Golden Melodram 20037 more expensive, less rich.
Medea, Scala 12/10/53: AV from BJR LPs. Warner is a re-EQ of OOP EMI 5 67909, itself a re-EQ of filtered Verona 27088/9. EMI and Verona had It/Eng libretti.
Medea, Dallas 11/6/58: Myto 00164 from CDRs of original tape, or AV if Myto is unavailable.
Medea, Covent Garden 6/30/59: The first release of the superior, recently discovered FM or BBC transcript disc source was Pristine PACO 119, not a label I like. In selecting it, be careful to choose mono only; avoid ‘ambient stereo’. AV offers the same-quality mono version, less expensively.
Medea, Scala 12/14/61: AV sensational and properly pitched; Myto 0292 a distant second, though also at correct pitch. All previous issues drifted down over a semitone by Callas’ entrance.

Anna Bolena, Scala 4/14/57: DVN-18, by far the best option. Keep It/Eng libretto from EMI sets.
Lucia di Lammermoor, Mexico 6/52: Opera d’Oro 1225 or whatever is cheapest.
Lucia Mad Scenes, Mexico 6/14 & 26/52: IDIS 6404 [‘Rarities by Maria Callas’]; OOP Memories HR 4581.
Lucia, Scala 1/18/54: AV (from IDIS 6419/20); Standing Room Only SRO-831-2; IDIS 6419/20; Incomplete and terrible sound, but sadly, the complete original tape was reportedly destroyed, leaving us with only this fragmentary version.
Lucia, Berlin 9/29/55: DVN-19 only. Warner is a re-EQ of EMI 5 66441 conflated with other sources.
Lucia, Naples 3/22/56: AV has 3/24/56 excerpts previously in OOP Myto 90319, but cut in available Myto 00296.
Lucia, MET 12/8/56: DVN-26, the most accurate original version, complete with Milton Cross and performer announcements. AV is also fine, the re-EQed sound perhaps first used for the 2007 Met rebroadcast.
Lucia, Rome 6/26/57: Complete on OOP Arkadia CDMP 422.2; in crisper sound on Myto 00133 but with an accidental cut of the orchestral introduction to the first vocal piece. Contact AV who has corrected it.
Poliuto, Scala 12/7/60: DVN-24 from the tapes. There is nothing even remotely as good. Keep the It/Eng libretto from OOP EMI 5 65448 or OOP Verona 28003/4 (which EMI copied and re-EQed, even stealing the translated libretto).

Andrea Chenier, Scala 1/8/55: bad source. Warner corrected the pitch (EMI re-EQ of Rodolphe sounded better).

Alceste, Scala 4/4/54: poor sound source, but Golden Melodram 2.0019 is most honest. Warner boosts upper bass, but is okay and very inexpensive. Neither have libretti. OOP Melodram 26026 had Italian-only libretto; AV may issue a libretto in electronic format.
Ifigenia in Tauride, Scala 6/1/57: DVN-11 (even older DVN version is wonderful). Warner comes from a multi-generational source away. OOP EMI 5 65451 was terrible, but keep its It/Eng libretto.

Tosca, Mexico 6/8/50: AV (from Melodram 36052, CD 1-2).
Tosca, Rio de Janeiro 9/24/51 excerpts: AV from Melodram 36052 CD-3, but far cleaner, surface clicks removed manually.
Tosca, Mexico 7/1/52: AV (from OOP G.O.P. 714-CD 2), Opera d’Oro 1248 (same master); Myto 00236.
Tosca, Act 2 excerpt, Ed Sullivan Show 11/25/56: in Gala 100.515 or Gala 316. Full video on YouTube.
Tosca, Act 2 in 12/19/58 Paris Gala: audio, Gala 324; Myto 00268; OOP EMI 5 67916. Video on EMI DVD, Warner BluRay (not HD); earlier EMI VHS visually less tampered; private VHS versions untampered.
Tosca, Covent Garden 1/24/64: Warner is most current and similar to (doubtless re-EQed from) EMI 5 62675. Also, Opera d’Oro 1142 and long-OOP Virtuoso 2697242. Warner is cheapest. None sound good.
Tosca, Act 2 Covent Garden 2/9/64: EMI DVD or Warner (non-HD) BluRay.
Tosca, Paris 3/3/65: AV (in-house pirate).
Tosca, MET 3/19 & 25/65: AV for both performances: Corelli and Tucker (in-house pirates).
Tosca, Covent Garden 7/5/65: DVN-2.
Turandot, Buenos Aires 5?-6?-7?/49, 2-minute excerpt: DVN-12.

Armida, 4/26/52: DVN-16 only. Myto 00155 copied it, altered the sound, and has no supporting materials. Warner re-EQed Myto, removed original ambience, and cut 12 minutes that DVN rediscovered.
Barbiere di Siviglia, 2/16/56: AV, correct pitch, better sound, 51 tracks; Profil/Hanssler 8015 was first (pressed) version at correct pitch but dull and 19 (often mislabeled) tracks. Other CDs a semitone sharp.
Turco in Italia, ‘Non si dà follia maggiore’ Rome 10/19/50, all that survives: Tima Club CLAMA CD48; also, somewhat improved by AV as a bonus to EMI-HMV re-channeled stereo LP Turco in Italia.

La Vestale, Scala 12/7/54: Membran 223253-311, German pressing best. Original tape very overloaded.

Aida, Mexico 5/30/50: AV (same as Melodram 26009 or IDIS 343/4).
Aida, 6/3/50 excerpts: I am only aware of IDIS 6404 [‘Rarities by Maria Callas’].
Aida, Rome 10/2/50 excerpts: AV offers this as an appendix to 1951 Mexico Traviata from OOP Melodram 26019; Archipel ARPCD 0010 but beware of Archipel’s 2-second silent gaps between tracks.
Aida, Mexico 7/3/51: AV from BJR LPs is thrilling; Myto 00150, also BJR LP transfer from Legendary, is smoother. Avoid Warner, a re-EQ of EMI 5 62678 with bridge from ironically better source to remove false applause between Act 2, sc 1-2. EMI was itself filtered Melodram 26015. Announcer cut.
Aida, Covent Garden 6/10/53: Testament SBT2 1355 (all that’s available).
Ballo in maschera, Scala 12/7/57: Myto 00131 is crisp, exciting, though the source change mid Act 2 is more noticeable, while AV transfer from BJR LPs is more even; OOP Golden Melodram 2.0009 is bland.
Macbeth, Scala 12/7/52: AV in two versions; I prefer corrected non-BJR source; BJR transfer also excellent; Myto 00310, like Nuova Era 2202/3 from which it’s sonically improved, has splices and a bad break between CDs; Warner is re-EQed Nuova Era with the disc break corrected but splices unaddressed.
Nabucco, Naples 12/20/49: AV best, using more of the source closest to the original; Warner re-EQs more of Golden Melodram 2.0010, itself fair.
Rigoletto, Mexico 6/17/52: AV excellent, directly from BJR LPs; Myto 00220 from BJR via Legendary.
Traviata, Mexico 7/17/51: DVN-27 only. After fifty years since its first appearance on LP in horrible condition, finally a worthy version of this great performance from the original acetates, also containing the extant excerpts from the performance of 7/21/51.
Traviata, Mexico 6/3/52: AV directly from BJR LPs only; Myto 00218 from BJR via Rodolphe cuts “Addio del passato” from the 1951 Traviata excerpts.
Traviata, Scala 5/28/55: AV best. There are two original sources: in one, the first half is in good sound; the other has mediocre sound throughout. Most releases use complete mediocre tape. AV and a few OOP others in the past (Cetra Opera Live LP LO-28, Foyer 2CF 2001) used the good-sounding first half.
Traviata, Scala 1/19/56: Myto 00210
Traviata, Lisbon 3/27/58: AV. Radio Portugal (RDP) recorded this but issued CDs from a CrO2 tape copy played at the wrong equalization so it sounds shrill. There have been several attempted corrections: the best may be Pearl 0228. A copy was made by the house for Alfredo Kraus, who gave it to the late Ed Rosen for release. It has no EQ problems. Rosen later sold it to EMI which issued it on LP in 1980 (HMV RLS-757, Angel ZBX-3910). The first EMI CD, 7 49187 (1987/1993), is very pleasant, if a bit filtered. EMI 1997 and Warner re-EQs are terrible. AV has issued one predating EMI, pitch-corrected, and includes Callas’ greeting. AV and OOP EMI 7 49187 are the best options.
Traviata, Covent Garden 6/20/58: AV, the only one with all the notes (!) and extant BBC announcements.
Trovatore, Mexico 6/20/50: DVN-25 is the only version that should be considered.
Trovatore 6/27/50, excerpts: Eklipse CD14.
Trovatore, Naples 1/27/51: AV (same as OOP Melodram 26001).
Trovatore, Scala 2/53: Myto 00314, finally at correct pitch (earlier Myto versions had terrible speed drift, as did BJR/Robin Hood RHR 500-C LP).
Vespri Siciliani, Florence 5/26/51: Testament SBT2-1416 on 2 CDs. House tape made for Walter Legge as an audition for Callas. Excellent sound, but overture was not recorded. A broadcast tape, in far inferior sound, has the overture. AV on 3 CDs retains Testament’s sound and includes the Overture and a ballet number not on Testament, restores some applause, and corrects a few glitches. Warner conflates the good source with the mediocre one, but in trying to equalize the quality, degrades the Testament source.

Parsifal, Rome 11/20-21/50: Warner from Cetra acetate discs they own with Kundry’s missing Act I lines added from inferior broadcast tapes. AV will be doing a version as well.

Live Recitals

Again, please note that I may recommend OOP versions–e.g., Gala, Eklipse, but list others as well. Warner has not issued any recitals as of this writing, and EMI’s efforts were generally poor (probably those Warner will eventually issue).

Major Bowes Amateur Hour,New York, 4/7/35: Madama Butterfly: ‘Un bel dì’–‘Nina Foresti’: DVN-1.
RAI Turin 3/12/51: DVN-1 (complete extant material).
RAI Turin 2/18/52, San Remo 12/27/54, Milan 9/27/56: free download from DVN of BJR 143 from RAI acetates; Gala 100.515.
Athens 8/5/57: Gala 316; Myto 00264; Forza and Tristan arias on DVN free download of BJR 143.
Dallas rehearsal 11/20/57: AV minimally processed; Gala 323 muffled, also Myto 0254.
Norma ‘Casta diva’ Rome 12/31/57: in DVN-1. Video also available from DVN.
Los Angeles 11/29/58: VAI 1182.
London-BBC 6/17/58 Pritchard cond., Tosca, BarbiereAV unprocessed, Gala 321 filtered.
London-BBC 9/23/58 Pritchard cond., Norma, ButterflyAV unprocessed, Gala 321 filtered.
Paris Gala 12/19/58: audio, Gala 324; Myto 00268; EMI 5 67916. Video on EMI DVD; Warner BluRay (not HD though); earlier EMI VHS less tampered and private VHS versions untampered visually.
Hamburg 5/15/59: audio, Gala 325; Myto 00280; video, EMI DVD (with Hamburg 3/62) or Warner BR (not HD). Earlier version on Sony VHS had more bows and applause.
Stuttgart 5/19/59: Palladio 4188; Myto 00277, EMI 5 62682; OOP Melodram 16534 retains the two instrumental sinfonias but appears to be a clicky LP transfer.
Amsterdam 7/11/59: Oddly, best on 1997 EMI 5 72030 (the rest of the 2-CD set is poor). There is a later EMI version in worse sound, 5 62683.
London 9/23/59 Macbeth, Pirata: DVN-1.
London 10/3/59 Bohème and Mefistofele arias: DVN download of BJR 143; AV CD–same as BJR 143, Gala 316.
London 5/30/61 Sargent, piano. Don Carlo, Le Cid, Mefistofele: AV, Gala 322 filtered, repeats Mefistofele 2nd verse in place of unrecorded 1st verse.
London 2/27/62 Oberon, Cid, Cenerentola, Macbeth Anna Bolena: Movieplay 100344, OOP Melodram 36513 has Bolena cabaletta missing from Movieplay. Bolena also on AV with the 1958-61 London arias.
Hamburg 3/16/62: Gala 322; video on EMI DVD (with Hamburg 5/59), Warner Blu Ray (not HD).
New York-Madison Square Garden 5/19/62 Carmen arias: DVN-4.
London 11/4/62 Don Carlo, Carmen: Gala 322 (year given inaccurately as 1964). EMI video: VHS incorrectly lists 9/62; DVD correctly dated; on VHS and DVD (and Warner BR) with 2/64 Tosca.
Berlin 5/17/63 Semiramide, Nabucco, Bohème, Butterfly: AV (& Stuttgart, Copenhagen); OOP Eklipse P-13.
Stuttgart 5/23/63 Semiramide, Norma, Nabucco, Bohème, Butterfly: AV (& Berlin, Copenhagen); OOP Eklipse P-13; Norma bonus in AV Paris 6/5/63 concert.
London 5/31/63, same program: DVN-4.
Paris (Théâtre des Champs-Élysées) 6/5/63, similar program, includes Cenerentola, Manon, Werther, Gianni Schicchi, no Norma: AV in original stereo–2-CD complete & bonuses or 1-CD Callas arias only & bonuses; Gala 321 arias, crisp mono.
Copenhagen 6/9/63 Norma, Nabucco, Bohème, Butterfly: AV correct pitch (& Berlin, Stuttgart); Gala/Movieplay CED-100345 is a semitone flat.  Semiramide, Gianni Schicchi not recorded/survived/included.

Much of this material was issued on EMI CDs from the poor sources that started to appear in 2003; compiled in a 2006 boxed set The Live Recitals, 3 68022 (new catalog nos. 3 68023-32)

Master Callas at Juilliard 1971-2: DVN-13 & 14; selections in OOP EMI/Angel 7 49600 (stereo).
1973-4 recitals with di Stefano. Audio and video on various labels, and in documentaries.


There are brief interviews and greetings before and in-between acts of complete live operas, like 1952 Traviata (AV, Myto); Met Lucia 1956 (AV); Lisbon Traviata 1958 (AV, RDP, Myto). Other short interviews and fragments can be found on DVN-1 and 4. All interviews are too numerous to document here.

Interviews with Mr. Rodrini, Milan 9/57; Norman Ross, Chicago 11/57; Edward R. Murrow CBS-TV 1/4, 11 & 24/58; Harry Fleetwood MET 3/13 & 27 1958; David Holmes, BBC 9/23/58: Gala/Movieplay CED 100341 in multi-volume ‘Callas Edition’. Video of Murrow interview in documentaries and various DVDs; easily seen on YouTube. Audio of above can also be located on YouTube.
5/25/63 with Michel Glotz (in French): bonus to AV 6/5/63 Paris concert.
6/4/64 with Bernard Gavoty ‘Le monde de la musique’ EMI DVD Classic Archive 46 ‘The Callas Conversations Vol. 2’.
4/10/65 Met intermission with William Weaver, recorded in Paris, late 3/65 or early 4/65. Bonus feature in both AV Toscas, MET 3/19 & 25/65.
5/18/65 with Bernard Gavoty; intermission of RTF studio concert: video EMI DVD Classic Archive 27 ‘The Callas Conversations Vol. 1’.
12/30/67, 1/13/68 with Edward Downes during MET broadcast intermissions: EMI 5 65822 or 3 68023.
Lord Harewood Conversations: 4/68, video EMI DVD Classic Archive 27 ‘The Callas Conversations Vol. 1’; bits edited out in audio on Movieplay.
9/13/68 with John Ardoin ‘Collector’s Corner’ Eklipse CD-33.
4/20/69 with Pierre Desgraupes ‘L’invitée du dimanche,’ video EMI DVD ‘The Callas Conversations Vol. 2.
Complete interviews from 1970-74 in various discs, also on YouTube. Fragments in various Callas documentaries (Ardoin, Palmer, etc.).

Studio Recordings

For the studio material, the task has become more complex with Warner Classics, the new owner of Cetra and EMI, having issued its 2014 set of all Callas’ commercial recordings (except two takes of 1949 Cetra Arias, discussed below). Warner remastered the ‘original analog tapes’ with 24-bit/96 kHz technology.

Having listened to the Warners for several years, either the engineers were very inconsistent or they were not being truthful about their sources. To begin, 24/96 technology, no matter how highly touted, is no guarantee of better results for analog material; often it is worse: harsh, boomy, thin. Moreover, CDs are encoded at 16-bit/44.1 kHz so a 24/96 transfer, of necessity, is down-converted for actual CD use.

Another problem is the putative ‘original analog tapes’. Especially with recordings made nearly seventy years ago in post-war Italy on brittle acetate-backed tape, what, exactly, are the ‘original analog tapes’? Obviously, the oxide on which the sound patterns were encoded, flaked off the backing many decades ago–the nature of acetate tape. Further, these ‘original tapes’ would have been copied numerous times using analog methods at various stages for international branches of EMI and Cetra to create master LP stampers. ‘Original tapes’ is therefore probably an impossibility, certainly for the earlier mono recordings.

A key issue is the unlikelihood of obtaining accurate digital results from old analog sources. The tapes were raw, not intended for listening but for processing via several analog stages. It was the final product that was judged and approved by the artists and producers. Those intermediate steps are not available via digital means, explaining the harsh or muffled results obtained by the digital manipulation of Callas’ mono material, the core of her output. It is most telling that all LPs sound plausible–except, perhaps, artificial stereo, or certainly, deceptively marketed LPs made from digital masters. Given its parameters, assuming the engineer had good materials, good information, and a good ear, analog mastering was always acceptable, even in the budget US Seraphim pressings from Capitol/EMI. No digital version of the mono material is accurate, and much of it is unappealing. Any mono LP will represent Callas nearly as she sounded. No studio CD truly can. Therefore, my CD preferences, in discussing the studio material, come with the qualification that I am choosing the lesser evil. If you have the LPs, listen to them.

Knowing that the availability of LPs and good equipment on which to play them is limited, I focus on CD options. There have been five official CD versions of the EMI studio recordings; four are differently equalized versions of the first analog-to-digital transfers: [1] pressings from 1984-90 with the EMI/Angel logo; [2] from 1991-95, as ‘EMI Classics’. Ostensibly from the same digital masters, with the same texts, look, track divisions, and catalog numbers as [1], these sound quite different; [3] the ‘Callas Edition’ (CE) 1997 (with live releases extending to 2003); EU and US pressings of this series, again, sound noticeably different: EU richer, US brighter. EMI apparently stopped pressing in the US during the latter part of this series. [4] Great Recordings of the Century (GROTC) 2002-5; [5] Warner Classics 2014, supposedly new transfers from analog tapes. [3] and [4] have been issued by EMI at differing price points, like slim-line budget ‘Historical Series’ reissues of CE or GROTC discs. Although LP transfers would seem like an excellent option to recapture the original sound, no efforts have yet been successful; e.g., Naxos, where fast CEDAR noise reduction ruined the sound in all but one case. The Cetra material, in the public domain for decades, has been subjected to countless editions. I focus only on the original Cetra CDs, 1984-7; the 1990s Cetras added reverb and other infelicities.

Of course, the easiest option in print is Warner, which I sometimes recommend below. But the other options are usually available in considerable quantities, used or new, and often for very low prices.

Complete operas

Norma 1954: CE 5 56271, EU pressing preferred but not essential: the 1997 Callas Edition which I generally dislike. However, an entirely satisfactory CD version does not exist. 1985/93 EMI 7 447304 (US 47303) is muffled; 2003 GROTC is very harsh, accentuating the overload break-up in the tape–which the LPs scarcely do; Warner is apparently a clone of GROTC with the mistaken 5-second silence before ‘Ma di, l’amato giovane’ finally removed (an LP side break, about which I had written often).
Norma 1960: Ironically, I again opt for the Callas Edition, CE 5 66428, US pressing only; the 1989 EMI transfer, 7 63000, is too close and oddly harsh, and the EU CE is actually worse! Warner is all right but it inserts more of the distracting reverb track that spoiled the original LPs (Columbia and Angel); the same erroneous 5-second silence is deleted here as in Norma 1954. EU CE has noticeable reverb as well, nearly undetectable in US CE. With no added reverb, the best version of all is the 1980 EMI-HMV LP set SLS-5186. The stereo recordings are generally better served on CD than the mono, perhaps because the extant tapes are closer to the originals, or that analog-digital processing is less harmful to stereo material.
1986 EMI/Angel 7 47308, West German pressing. 1992 EMI Classics pressing is smoother but has less rich vocal core. CE EMI 5 56275, EU pressing not bad, if leaner. Warner is poor, robbing Callas of the warmth in her heavy voice and over-accentuating/over-etching the orchestra, especially the bass, at the expense of accuracy and voices (a trend with Warner).
Sonnambula: No CD recommendation. A vivid demonstration of why raw analog 1950s studio tape transferred to the digital medium is problematic. All CD versions are absurdly boomy and present the Callas’ Amina lacking much of the sweetness and gentleness that we know from Callas’ live performances and LPs of this set. Perhaps Warner is marginally less inaccurate tonally. Listen to the LPs.

Carmen: 3-CD 1985 EMI/Angel 7 47313, Japanese or UK pressing; 2-CD Warner. The 2-CD 1991 EMI Classics 7 54368 is muted; CE 5 56281 is hard and lean.

Medea: CE 5 66435, EU pressing only, which, while unlike the LP, is exciting. Warner, having lost the rights to the tapes (claiming they ‘couldn’t find’ them), is a dull re-EQ of 1990 EMI 7 63625; the actual set, West German pressing, is close while the EMI Classics, Dutch pressing, is more spacious.

Lucia 1953: 1989 EMI/Angel 7 69980 is the default choice, with the richest sound and great presence, but it has distortion not in any LP version. CE 5 66438 removes the distortion but also the vocal immediacy and complexity. Warner, at first, seems most like the LPs; but Callas’ heavy voice is thinned (it is also thinned on some LPs, but not that much), and the digital glare, especially on high notes, can be ear-splitting at times. Like Norma 1954, it is virtually identical to the GROTC version (2004), 5 62747.
Lucia 1959: 1986 EMI/Angel 7 47440, Japanese pressing, gives Callas’ voice so much presence that its peculiarities at this period are fascinatingly revealed. The later EMI Classics pressing is kinder but less riveting (the US pressing also has a new electronic click in all copies), and CE 5 56284 pushes her farther back. Warner splits the difference between the first and later versions, and is cleaner, though, as always with Warner, digital glare is a factor. The best version is the Japanese LP EAC 47137-8.

Pagliacci: Warner is the default, the only one without added reverb. But there is a caveat. Despite Warner’s claims of authenticity, it seems more that they simply removed the added reverb from the digital tapes, as it doesn’t reproduce the LP’s incisiveness, especially the amazing red-label Angel ‘Made in England’ or Columbia LP, which leaves the listener breathless from its almost obscene realism.

Cavalleria Rusticana: Overload distortion permeates this recording, though, like Norma 1954, it is scarcely a problem on LP. On CD, I prefer CE 5 56287 EU pressing, or its budget-priced clone, 2005 Historical Series 5 86830, simply because the older 1987/93 version breaks the opera over 1.5 discs (CD lengths were only 74 minutes in the 1980s). Warner is bass-heavy and oddly recessed. Sadly, with the EMI versions, one must take the reverbed Pagliacci, as the two operas are packaged together.

La Gioconda Cetra 1952: OOP 1986 Fonit-Cetra CDC-9. Warner 2014 is the most available and removes the reverb afflicting decades of public domain issuances, including Cetra’s 1990s CDs and clones (e.g., Hommage). But Warner compromises and chills the golden richness of Callas’ heavy voice. Warner’s jacket picture is from the 1960s, ‘Meneghini’ omitted from Callas’ name.
La Gioconda 1959: 1987 EMI/Angel 7 49518 West German pressing, has amazing presence. CE 5 56291 is muffled. Warner is lean and has the usual digital glare. The original LP was not good.

La Bohème: 1993 EMI Classics 7 47475, US 1993 pressing only. The 1987 EMI/Angel pressing is muffled. That a later pressing ostensibly from the same master could be so much better has always been a puzzle. CE 5 56295 is terrible. So is Warner, where the voices sound right but are swamped by a bass-forward orchestral re-EQ, ruining the intimacy of this most intimate of operas.
Madama Butterfly: Same comments apply. 1993 EMI Classics 7 47959 US pressing only is acceptable, not 1987 EMI/Angel. CE 5 56298 and Warner are similar: thin and unappealing.
Manon Lescaut: 1985 EMI/Angel 7 47393 West German pressing. Interestingly, even the simultaneously issued Japanese pressing lacks the warmth and beauty of this one and has some digital glare, as do the 1992 EMI Classics pressings. CE 5 56301 is thin. As for Warner, the comments for Bohème apply.
Tosca 1953: Warner; CE 5 56304 EU pressing, revised 1999. Among the greatest operatic recordings, even sonically, Tosca’s post-analog life has been hard. First digitized in 1984 as Angel BLX-3508 even before its CD issuance–‘X’, signifying ‘digitally remastered’, and red replacing black on the album cover, I remember my shock upon hearing it: the top two octaves muffled, exciting brass ‘fizz’ at Scarpia’s murder gone, and Callas sounding veiled. On CD, 7 47175 (US 47174), it was issued along with the same digital LP version, now budget-priced in the ‘Angel Voices’ series, AVB-34047 (EMI-HMV logo and ‘Angel Voices’ deleted in the EU). The 1992 US EMI Classics CD is its best incarnation, a bit crisper and brighter, but still disappointing. The 1997 CE re-EQ had a glaring error: Tosca’s three opening calls of ‘Mario’ were moved closer together, deforming Callas’ interpretation, the third call replaced with a copy of the second. I got EMI and engineer Allan Ramsay (later, head engineer on the Warner edition) to correct this terrible edit for another pressing run. With the correction, it improves the soggy 1984 transfer (though the mistakenly reduced volume of the third ‘Mario’ was retained), restoring ‘fizz’ and better vocal clarity, though, like most of CE, it does recess the voices, though not too much here. The catalog number never gave any indication so one can’t know whether one has the corrected version in hand. 2002 GROTC 5 67756 (US 5 67759) is the result of a bad miscalculation: Ramsay (again) ‘solved’ perceived speed drift by starting at a lower pitch, a=435, so the recording ends at standard a=440, adding 1.5 minutes to a 108-minute opera. Warner has some of the LPs’ voluptuous, focused sound that ‘lifts the veil’ as one critic put it; it corrects the volume of ‘Mario’ no. 3; but after subtracting extra lead-in and lead-out seconds, it still runs almost a half-minute longer than versions other than GROTC, though the pitch appears to be correct.
Tosca 1964CE 5 66444, US pressing only. The 1989 EMI/Angel CD version, 7 69974, is too close and airless, the damage in Callas’ voice highlighted; and as with Norma 1960, the EU CE is even worse. Warner is less kind to Callas than US CE, but not as cruel as EMI 1989 or EU CE. US CE is preferred, presenting Callas’ voice more attractively, and more similar to the LPs.
Turandot: 1987 EMI/Angel 7 47971, West German pressing. The first LP was superb, as was the (probably still available used) 1980 LP UK EMI-HMV RLS-741. The CDs typify the good and bad points of digitally transferred mono Callas. The 1987 West German pressing has a huge, close presence, but it is airless; the 1993 pressing is airless as well as muffled. CE 5 56307 has fascinating instrumentals while Callas is further-muted as well as recessed. Warner is closest to the balance heard on the original LPs but lacks their color and core of vocal sound.

Il barbiere di Siviglia: 1986 EMI/Angel 7 47634, West German pressing, the version to which I return. It is an instance where the early stereo LPs are poor, like all their faithful Angel successors. The best LP is UK EMI-HMV SLS-853. The 1993 EMI Classics CD pressing loses the 1986 rich presence even though it’s a bit more transparent. CE 5 56310 is thin, wiry, and recessed. Warner is crisp, detailed, and clean but it places the listener at arm’s length from the humor and the humanity in the voices. It also reproduces the US cover rather than the UK design used for the rest of the series, and like Warner’s Cetra Gioconda, a post-1959 US cover at that, with ‘Meneghini’ missing.
Il Turco in Italia: 1993 EMI Classics 7 49344, German pressing. Like Bohème and Butterfly, an instance in which the 1987 EMI/Angel pressing is heavy and muffled while the 1993 German pressing is inexplicably far more transparent and accurate. CE 5 56313 and Warner are thin; Warner additionally has some odd distortion issues.

Aida: 1987 EMI/Angel 7 49030 West German pressing on three discs has warmth, presence, and color. 1992 EMI Classics is also good. CE 5 56316 is one of the worst CE examples: muffled, electronic-sounding, and vocally recessed. Warner is pretty good in replicating LP vocal size if not quite the presence, on two discs, all of Act 2 on the first disc–two seconds shy of the maximum 80-minute length.
Un ballo in maschera: 1987 EMI/Angel 7 47498 is my preference but this is a rare instance where every CD edition has virtues, though Warner is mastered too loud (they all are).
La forza del destino: 1987 EMI/Angel 7 47581 West German pressing is, in some ways, more satisfying than the LP. For one thing, Act IV, scene 1 is included, everything recorded finally present. In order to fit the opera onto three LPs, the scene had been cut in every release except a late 1970s UK LP set in re-channeled stereo where second verses of two arias were cut instead. The later US CD pressing, also ‘EMI/Angel’ rather than ‘EMI Classics’, changed nothing and is also excellent. CE 5 56323, like Aida, is recessed and muted, and Warner is virtually identical to CE, even timing exactly the same after the additional seconds between acts and lead-in/lead-outs are subtracted. Many mono Warners fit this description, a key reason why it is seems plausible that at least some are re-EQs of CE or GROTC rather than new transfers from analog tapes where such exact timings are very unlikely upon replaying, given the issue of tape tension changes from one end of a reel to the other.
Rigoletto: 1986 EMI/Angel 7 47469 UK pressing (West German is warmer, but UK is better balanced); EMI Classics pressings tend to be thinner. CE 5 56327 has an editing mistake, depriving Callas of the sibilant ‘s’ in sarà’ on the climactic note of the cadenza to ‘Caro nome’; unsurprisingly as we will see, the engineer responsible was Paul Baily. Warner is cold.
La Traviata Cetra 1953: 2005 Naxos 8.110300-01, the only Naxos worth having, if difficult to find: a crisp, warm, open LP transfer by Ward Marston. 1984 Fonit Cetra CDC-2, an OOP early CD from the tapes, lacked track divisions except at scenes and acts, making it cumbersome to use. But AV has just reissued it with 50 tracks, making it a good option. Warner has similar problems as their Cetra Gioconda: Callas’ rich heavy voice is robbed of size and warmth; it also seems recessed.
Il Trovatore: CE 5 56333 US pressing only. Given that all LP versions sound glorious, the troubled life of this recording on CD classes it with Tosca 1953. The 1987/93 EMI 7 49347 pressings and EU CE are heavy and muffled. But the usually inferior US Callas Edition pressing is excellent: crisp and nearly accurate. That was my choice in 2005 and remains so, though it now may be difficult to find. The 2004 GROTC 5 62898 dulled the sound again, though not nearly like 1987/93; and Warner, while presenting the voices relatively accurately, overbalances them orchestrally. GROTC and Warner are second choices.

In 2007, EMI issued The Complete Recordings, 3 95918, a 70-disc box. including the Cetra material (presumably from the reverbed versions then available), the complete EMI operas chosen from CE and GROTC, whichever was more recent, and all the studio recital discs as well (discussed below).

Studio Recitals

Cetra Arias 1949: 1999 Italian Warner-Fonit 8573 80575 ‘Il primo disco’. Although Warner is self-congratulatory about its 2014 version, their use of CEDAR, while creating a silent ambience, filters out the few valued high frequencies and space, already minimal on a 78. Both masterings use the version of ‘O rendetemi…Qui la voce’ from Cetra 78 CB-20483, where ‘O rendetemi’ ends with a glottal stop from Callas and a silence before the orchestra begins ‘Qui la voce’. This was apparently the version most issued on 78 but not on LP. As the music between sections is cut, it is difficult to know which Callas preferred, but the version with a smooth elision is Cetra 78 CB 20477, used for every LP release. It can be heard on Warner Fonit 8573 82241, though the whole is in less articulate sound than ‘Il primo disco’. A discarded take of ‘Casta diva’s first verse, recorded on 11/8/49, is (or was) in IDIS 6404; it is now included as a bonus in AV London, Covent Garden 11/18/52 Norma, in somewhat clearer sound.

EMI Studio Recitals. These too have had a difficult life on CD. In the 1980s, CDs were expensive and there were few pressing plants, so no mid-price option existed. EMI therefore felt obliged to provide discs with over 70 minutes of music. Sadly, the very well-considered content of Callas’ 39-49-minute recital LPs was not considered adequate, so EMI created 70+-minute CDs culled from various LPs, with very odd results. The 1958 Mad Scenes now included the Donizetti arias from the 1964 Rossini/Donizetti LP as well as the entrance aria from Bellini’s Il pirata, recorded in 1961 and released on the 1972 Callas by Request. The resulting CD was called Mad Scenes and Bel Canto Arias, 7 47283. The 1958 Verdi Heroines, with material appended from 1964 material issued on Callas by Request, became Verdi I, 7 47730. The 1964 LP Verdi Arias was expanded on CD to include unapproved material from 1964 and 1969 and called Verdi II, 7 47943. The 1961 Maria Callas Sings Great Arias from French Opera now included most of the 1963 Maria Callas in Paris: Great Arias from French Opera, Vol. 2; the CD is known as Callas à Paris, 7 49059. And a disc generically called Opera[tic] Arias, 7 49005, was a hodgepodge of the remaining 1963 French arias, the Rossini half of Rossini/Donizetti, and most of the Beethoven/Mozart/Weber LP, except for the Beethoven concert aria ‘Ah! perfido’, which was issued as filler on the 1990 Medea, 7 63625. As peculiarly assembled were the CDs that had never been LPs: the 1987 Maria Callas: the Unknown Recordings, 7 49428, and the 1992 Maria Callas Rarities, 7 54437, both of which combined unapproved studio material and excerpts of live 1950s RAI recitals from poor sources; Maria Callas Rarities also repeats the Beethoven concert aria ‘Ah! perfido’ already on the complete Medea. The only conflation CDs that made any sense chronologically were the two earliest: Puccini and Bellini Arias, 7 47966:the 1954 Puccini Heroines with the 1955 Sonnambula arias originally intended for Callas at La Scala but unapproved and left unissued, as well as Opera[tic] Arias (same generic title as a disc already described), 7 47282:the 1954 Coloratura-Lyric LP with the approved Medea and Vestale arias from Callas at La Scala appended to it. In fact, this latter disc was itself remastered as a 2005 GROTC entry, 4 76843, misleadingly titled Lyric-Coloratura Arias, with a reproduction of the original UK LP cover for Maria Meneghini Callas Sings Operatic Arias.

The frustrating part of the first release is that the masterings are excellent, with Callas’ voice rich and forward (though the French ‘MPO’ pressing of Mad Scenes and Bel Canto Arias is shrill). The EMI/Angels are luscious while the later EMI Classics pressings, as expected, are more limpid and spacious. In the past, I reassembled the material into their original forms on CDRs for my own use. Beginning in 1996, EMI had the same idea: with CDs more easily and cheaply pressed, shorter mid-to-low-priced discs became an option.

The first Callas recital discs on CD with original content were also the best. Sadly long out of print, these were mastered by the brilliant Yoshio Okazaki and issued singly in 1995; in September 1996, they appeared as a boxed set unrevealingly entitled The Art of Maria Callas on Toshiba-EMI, TOCE-9166-9180. In the last decade of the LP, Japanese EMI Callas masterings ranked among the best, and this set of CDs continued that tradition. The discs accurately reflect the original LP content, including the last one Callas approved, Callas by Request, with Verdi arias recorded in 1964 and the Pirata entrance aria from a 1961 session. The set goes beyond that in reproducing the first LP issued after Callas’ death, Maria Callas the Legend: unapproved 1964 and 1969 Verdi arias and the 1955 Sonnambula arias’ first appearance–in rechanneled stereo as they had first appeared on LP (not quite a nod to authenticity I might have elected). The Act II aria from Verdi’s Il Corsaro, ‘Né sulla terra…Vola talor dal carcere’, is offered with its cabaletta, ‘Verrò…Ah, conforto è sol la speme’ that had been omitted from all but the first French LP release. There is one unsanctioned addition to the first French recital: Saint-Saëns, Samson et Delila, ‘Mon cœur s’ouvre à ta voix’, first issued on LP in 1982; Callas had adamantly refused to approve its release even after she had approved other previously unsanctioned material in 1972. The set reproduces the chaotic Maria Callas: the Unknown Recordings and Maria Callas Rarities in beautifully mastered sound. The last four discs present problems, even Callas’ approved By Request, because with them, the chronological sequence is broken. Nevertheless, Okazaki’s set is the one to cherish for its musicality, sonic accuracy and the belief in preserving historical precedent. It also includes a disc of Callas’ 1967-8 conversations with Edward Downes, originally a bonus LP in the 3-disc set ‘La Divina’, Angel SCB-3743.

UK EMI, as part of its 1997 Callas Edition, also released a series of shorter discs that largely reproduce the LPs at low prices, separately as 5 66457-67–catalog nos. not chronologically sequential, or in a box as 5 66618, reissued in 2006 as The Studio Recitals, 3 68033 [3 68034-46], or in 2007’s The Complete Recordings, 3 95918. But there are striking differences from the Toshiba-EMI set. First of all, the sound is often thin and recessed like much of the Callas Edition. Secondly, and more egregiously, new editing and pitch errors afflict many of the discs, dismissing them from consideration. Puccini Arias was mastered with the two identical mono channels accidentally out of phase with each other. Lyric-Coloratura (note title-word inversion) has an ill-considered edit in the first aria, removing Callas’ glottal cut-off of her final note. Both discs were mastered by Paul Baily, whose other editorial decisions, like deleting Callas’ sibilant ‘s’ in ‘sarà’ of ‘Caro nome’, Rigoletto CE 5 56327, as well as her behind-the-curtain vocalizing from his ICA 1958 Covent Garden Traviata, ruins them. Four discs–both French recitals, now called Callas à Paris I and II, Rossini/Donizetti, and Beethoven/Mozart/Weber–were mastered a quarter-tone sharp by Simon Gibson; aria order in Rossini/Donizetti was changed. Regarding unapproved material and the original order of the LPs beginning in 1972, EMI came to some odd decisions. A new CD called Verdi III, 5 66462, 3 68044, was created, containing both approved and unapproved material from 1964 and 1969. The unsanctioned 1955 Sonnambula arias (in original mono) were placed on Callas at La Scala, unlike the LP. These decisions retain the chronology of her studio work–with one bizarre exception: the sanctioned 1961 Pirata entrance aria first issued in 1972 on Callas by Request, is omitted from the principal CDs, appearing inappropriately in EMI Rarities, 5 66468, 3 68045/6, a release of all the unapproved material.

The 2014 Warner Callas Remastered reproduces the EMI recital discs with some unfortunate new problems. The three remastered mono discs are vocally even thinner and harder than CE–the worst I have heard them, imbalanced by over-present bass, though the 1997 editing and phase errors have been corrected. Although they initially give the impression of being vocally-sized like the LPs, that impression is dispelled in a few harsh, thin, cushion-free notes, a disservice to Callas. On a positive note, the authorized stereo recital discs 1958-64 sound good and have no pitch mistakes (though are mastered very loud like the entire set). The Corsaro cabaletta has been deleted from Verdi III and is now on Rarities, a completely inexplicable and pointless decision. Despite singing some notes mezza voce for the cabaletta, Callas never approved any of this scena, so if it is issued at all, splitting it up is merely inconvenient for the listener. Rarities sounds muddy compared with the EMI version.

In the book accompanying the complete Warner set, head engineer Allan Ramsay describes, with peculiar pride, his addition to the recital discs of inter-track background hiss resembling what he thought sounded like LP noise. The truth is that complete silence between bands was the hallmark of fine LP pressings, so ironically, Ramsay has emulated mediocre LPs. He does not mention that the CE recital discs were also treated in this fashion, as are the beginnings and endings of discs, acts, and scenes in the complete operas on Warner, CE, and GROTC. It is a very strange idea, and to my knowledge, has never been imposed upon CD remasterings of analog recordings by any other artist.

The choices for the EMI recital material, therefore, are rather complex.
[1] The 1996 Toshiba-EMI set The Art of Maria Callas, TOCE 9166-9180, is the best option, but one must ignore the chaotically assembled 1987 Unknown Recordings and 1992 Maria Callas Rarities CDs, replacing them with 1997 CE EMI Rarities, 5 66468, 3 68045/6, to complete it.
[2] The next option is a strange one: the Puccini/unapproved Bellini disc, EMI 7 47966, and Opera Arias EMI 7 47282–that is, Coloratura-Lyric/sanctioned Callas at La Scala material, with the 1958-64 stereo discs on Warner, but Verdi III on EMI 5 66462, 3 68044, and CE EMI Rarities, 5 66468, 3 68045/6, which sound better and are more logically assembled than they are on Warner.
[3] The third option is to assemble one’s own CDRs from the 1987-92 discs in their EMI/Angel or EMI Classics pressings (it’s not difficult), but one will need 1997 EMI Verdi III, 5 66462, 3 68044 for the Corsaro cabaletta, not issued on CD until 1996. And then, the 1997 CE EMI Rarities, 5 66468, 3 68045/6, is necessary for correct versions of unreleased studio material—and the misplaced Pirata entrance aria.

There is no single satisfactory option except the LPs–though again, one would need CE Rarities, 5 66468, 3 68045/6. My preferred release, with ideal sonics and surfaces, is the ca. 1983 11-LP set, French EMI 2C 165-54178/88: Maria Callas: Ses Récitals which has the same cover photo (1958 Paris Gala) and title lettering as the unappealing 1997 CE boxed set, EMI 5 66618. The LPs through 1964 are presented in their original forms with two exceptions: the unsanctioned Sonnambula arias and the 1961 Pirata entrance aria (an odd choice, but at least it’s among authorized material) in Callas at La Scala, and ‘Mon cœur s’ouvre à ta voix’ in the first French disc. The final disc in the set is the first appearance of what was later called Verdi III (no Corsaro cabaletta, however), culled from Callas by Request and Maria Callas the Legend. The set comes with a copy of ‘L’Avant Scène’ no. 44, devoted to Callas, with excellent articles, aria texts, and full documentation of her appearances and recordings.

1972-3 London sessions with Giuseppe di Stefano: DVN-20, great sound but John Ardoin’s observation of the Aida duet on this disc applies to the entire project: ‘it would make a stone weep’.


—Gratitude to M. Petkovic for editorial assistance. For questions, comments, corrections, etc. I may be reached at

©2019 Robert E. Seletsky