Christian Xell from Tangotunes.com asked me a couple of months ago to write a small article about Juan D’Arienzo’s second studio contract during the years 1935-1939. Here is an excerpt:

On 24th June 1935, Carlos Gardel, the most popular tango star at that time, died tragically. Such was Gardel’s prominence in Tango music that his sudden death left a void, but that was to change. Precisely 8 days after this sad event, Juan D’Arienzo reentered the recording studio, signing an initially short contract with the Victor label for a series of instrumental themes. This time, D’Arienzo would put aside his violin and take the lead as conductor for his ensemble. The first record he produced with Victor included the vals « Desde El Alma » on the A-side and the tango « Hotel Victoria » on the B-side. As it turned out, he would remain faithful to the same record label for the rest of his long career, laying the foundations that would revolutionise the tango genre forever, and open the door to the Golden Age of tango!

D’Arienzo constructed his new arrangements around four accented beats per bar, creating the impression of a more sustained rhythm where the beat is over the melody. The new « D’Arienzo sound » was characterised by a faster tempo that suggested more action and movement, it subconsciously recalled fundamentally natural rhythms, like that of a heartbeat. This new music had a great impact on all the major bands of the time and by the end of the 1930s most of the orchestras had adopted a faster pace and stronger accentuation of the beat. Everybody got caught up by the D’Arienzomania and tried to catch up with its success.

[…]

You can find my full article on the Tangotunes blog

Christian and his team were busy on making new transfers from shellacs available on their website which cover this complete period of D’Arienzo’s second outfit. Some of these titles had already been published previously by Tangotunes but they decided to do the whole series again. Right now three download volumes are finished and already available on their website but they want to release all of the titles. The rest will be available soon. I think it’s worth to dig a little deeper into this great restoration work as it has some aspects which witness very high transfer standards.

In recent times there have been two major papers which sustainably unveiled a new approach towards transferring and dealing with tango records. One is Age Akkerman’s manual on how to retune tango recordings called Lost in History – the Keys of Tango Music. This analysis is based on the conclusion that most of the CD transfers which are available nowadays are out of tune, most of the time overclocked. As a result all the instruments and voices in the recordings sound false and unnatural and he shows a method to bring them back into tune by analysing the frequencies of the A reference tone, which is also called the concert pitch.

Often these pitch shifts were introduced intentionally by some recording companies in later reeditions, see my previous posts on this subject, or, happened directly during the recording sessions. Peter Copeland shows in his Manual of Analogue Sound Restorations also some possible causes which could have influenced on the correct speed of a 78 rpm recording, like fluctuations in the power network which wasn’t yet so stable as it is nowadays. In the 1930’s, especially in cold winters, the frequency of the current could vary from the standard 60 Hz oscillation and make the motors of the record cutter run at a slightly different speed (nevertheless Buenos Aires is less likely to have had very cold winters) … Most of the record players at that time, as Age stressed out, had a pitch control, permitting for an instant correction by the listener.

The other paper is that of Igor “El Espejero” Shpigelman, called the Equalizer-4-TJ, The Equalizer for Tango DJs. It’s a comprehensive guide which shows how to use an equalizer to get the best sound out of a given tango recording. Showing precisely how the sound spectrum of a tango recording goes and where frequencies might be attenuated or amplified to obtain a certain result to fit the venue or the sound quality. This work is very interesting as it does not exclusively apply to the playback domain, the tango djing, but also shows some interesting findings which could be applied to the transfer of recordings.

According to Tangotunes, the idea of the new D’Arienzo series is to offload most of the processing into the analog domain. Using a good restoration preamplifier and an equalizer to have the best and purest deemphasized signal for a careful last digital processing mainly limited to click and crackle removal. If you put together the two before mentioned methods of retuning and equalizing and offload those steps into the analog field then you can get excellent transfers. And that’s what Tangotunes did!

Doing the retuning in the analog phase, by speeding up or down the turntable just inverses what happened when the recording speed fluctuated and the musical keys went out of tune. Doing it on a digital sound file, though, by computing all samples, might introduce quite some interpolation errors. Where information is missing, the computer will add something, so the result could be different. By the way, the computing gets more and more imprecise the higher the percentage factor of a digital pitch filter.

And indeed, up to now, the best transfers I can currently think of are earlier LP transfers done by the Japanese collectors label CTA, Club Tango Argentino. Their LP transfers are among the best restoration works on tango recordings I ever heard. To a certain extent, these vinyls sound even better than their recent CD transfers as they did most likely not undergo any digital processing at all. Hence, reproducing the initial sound signal of the original recording session in the purest imaginable manner.

It’s interesting to recall what these original recording sessions were: They were direct-to-disk recordings. No multichannel, no tape, no editing, no overdubbing, just plain recording directly onto the wax platter without any stops, intermediate or post-production steps. This also meant that there was no way to cheat or to alter the result and there had been very few takes because of the high price and intensive preparation work of the recording support. Most of the time these sessions were done in one flush at first take. Unimaginable today. Though, this direct-to-disk process had been around until the end of the LP era as you can read in the aforementioned Wikipedia article! And it’s coming back, look for instance at the new label Berliner Meister Schallplatten, they are reintroducing the direct-to-disk process with vinyls and taking it to a new climax, they even invite the public into the studio during a recording session. The direct-to-disk process makes me think about Roland Barthes who wrote about silver photography in his essay “La Chambre claire”, long before the advent of Photoshop, that the noema of photography is the shocklike ça a été (that has been), as with these direct-to-disk recordings which are an immediate conserved state of sound without any cosmetics! A shocking document of what has been … A great deal of the emotions we live while listening or dancing to these recordings certainly result from this circumstance!

The Tangotunes series has the mention “Golden Ear Edition” to distinguish it from their former transfers of the same orchestra. Now, the retuning effort clearly merits this self assigned label. It’s something quite special which hasn’t been done yet openly for larger reeditons. According to Tangotunes, they used the CTA D’Arienzo transfers (CTA-301 to CTA-318) as a reference and found out that during the year 1939 the average A concert pitch measured in these recordings suddenly changed from 435 Hz to 440 Hz:

“When we measure the pitch, it should be taken from the piano in the recording, provided that the tuning of bandoneons could range from 435 Hz to 445 Hz. The violins are not a well-tempered instrument and the musicians always adjust themselves to the piano or the bandoneon, so we cannot rely on them.

Since the only possible pitch would be 435 Hz and 440 Hz, the figures significantly show that:

1. The standard pitch of 435 Hz was used by D’Arienzo’s Orquesta until 1939.
2. The standard pitch of 440 Hz was applied by D’Arienzo’s Orquesta since 1940
3. The CTA recordings, statistically, were 2.60 Hz higher transferred in average than the original play, which is around 10-15 cents.

[…] Naturally, we can find that the time when D’Arienzo’s orquesta was retuned is between 1939-03-03 to 1939-04-18.”

Source: Frank Jin of Tangotunes

The dramatic pitch jump up to ~448 Hz in the diagram between 1939 and 1941 could also be a hint that the speed of the recordings could have been intentionally increased on some of D’Arienzo’s studio sessions at around 1940-1941 as the Tangotunes document further states. According to them there are also some recordings where one is able to hear differently tuned bandoneons playing together during this retuning periode, like in Di Sarli’s recording El retirao from 11.12.1939 …

I have traced the spectrum of different versions of the recording El cencerro, which is one of the titles of the new Tangotunes series, all are in 44.1 kHz 16 bit, FLAC, the last Tangotunes is in AIFF. What we can see in the following spectrograms is that the first, taken from the El Bandoneón CD-43, has nearly sheer edges on the left in the bass, cutting off at around 60 Hz and on the other extreme, a high cut starting almost at around 4 kHz, leaving not so much music in the low and high frequencies. In the middle range you can see also a drop between 1000 and 2000 Hz, maybe due to the strong echo effect which had been added to this version.

Overall spectrum of El cencerro from the El Bandoneon CD EBCD43

Overall spectrum of El cencerro on the Audio Park CD Vol. 1

Overall spectrum of El cencerro from the CTA-303 CD

Overall spectrum of El cencerro from the Tangotunes download, AIFF file D’Arienzo Vol. 1

Comparing the Audio Park to the CTA, one can see that Audio Park cuts off a little less towards the 20 kHz, in the treble, compared to the CTA version. The CTA on the other hand seems to leave in all the infra-bass region. Just judging from the spectrum the Tangotunes seems to have the most balanced frequency range. Listening to it confirms this very good impression …

One word about the background noise which is quite subtle on the Tangotunes AIFF version and a little stronger on the CTA and Audio Park CDs: Once coffee and milk are mixed, it’s difficult to remove the milk without removing some coffee too. The same applies to sound restoration. It’s better to attenuate the background noise and not to try to remove it completely. The El Bandoneón version is the quietest because of its strong shelve filters in the bass and treble regions but it’s therefore missing music too, sounds muffled and has added artefacts, a strong reverb effect and other noises resulting from heavy filtering and which are actually not in D’Arienzo’s music! From a psychoacoustic point of view, the human ear is very well capable of removing carefully attenuated background noise from our perception. Whereas it is very aware and annoyed by artefacts and missing frequencies. In the end, we can use an equalizer to cut off hiss and other background noises and individually decide how much to remove during playback. Most of the time, anyway, it’s the bad PAs at the venues which eat all the background noise as they tend to be good in bass and mid-range reproduction but often are deficient in high tones. From my experience, events with audiophile 3-way systems are rare.

As Tangotunes is dedicated to music for the modern milongas the two polcas from 1937 will be missing as polca is not so much in vogue anymore 🙂 By the way, the Tangotunes version of Don Esteban is, like the CTA version, take 2. Could be cool to add take 1 with the two added Biagi solos too which is quite rare without reverb effect. For a deeper analysis of take 1 and 2 see my article News from Don Esteban.

Their recent Ángel D’Agostino transfers were already very good but they missed the retuning part which is now integrated into this new series. Already since the D’Agostino series, their sound files look very well done, no clippings left as it had been the case with some files in former series.

During our talks, Christian told me: “What I find most exciting is that everybody can use their ears to get an idea if a recording sounds right and if the speed is correct. Especially with strong speed fluctuations this is easily possible.” So let’s trust our ears, that’s what it’s all about!

Since quite recently I observe a general shift in tango djing. A lot of DJs are today putting a lot of effort into recovering the most authentic sound in the milongas. This translates into investing in high end sound devices, prefering lossless sound formats over lossy codecs, analysing sound files and playing correctly tuned tracks. Trying to find tango tracks in unfiltered versions and without certain artefacts which have been added to the original recordings in later copy cycles. Unveiling somehow unplugged, purer versions of the presented repertoire with a close to perfect sound. They rather like to have some surface noise than a cleaned to death recording!

To visualise this effort, one has to be aware that the time interval which is of most concern to modern dancers, 1925-1955, the Epoca de Oro with pre and afterparty, is parallel to the technical invention of the electrical recording systems in the second part of the 1920s and ends with the introduction of the modern vinyl record and the first stereo systems around the mid-1950s when the full frequency-range recording had been finally made possible.

Westrex electrical cutting

If we look at the records’s frequency spectrum we can see that in the beginning certain frequencies especially in the mid and high ranges weren’t yet completely present and towards the 1950s these frequencies were successively added as the recording systems improved. Let’s take the year 1926, most of the record companies licenced the new Westrex system and were able to reproduce a recording bandwidth from 50 Hertz to about 6,000 Hertz, beyond which the high frequency sensitivity declined. This had been a sensational improvement compared to the former acoustic recording systems (200 to 2,400 Hertz and sometimes less) which were in place until around 1927 mostly in an effort to empty their stock of acoustic records. None of the older acoustic recordings (nearly everything before 1927) are played in any milongas nowadays. They don’t suit our needs because they are close to inaudible and one gets easily tired from them. But listen to a very early electrical Argentinian recording, A media luz interpreted by the Orquestra Francisco Canaro from around November 1926. Though from the early electrical recording days the sound quality is astonishingly good and it still fits perfectly into a typical today’s milonga program.

• Early electrical recording: A media Luz Francisco Canaro : Instrumental Tango 1926 3:05

• And compare it to this acoustic recording just from the year before: Mesalina Francisco Canaro : Instrumental Fox-trot 1925 2:10

The missing frequencies let the tangos, we like so much, sound to our contemporary ear like if they were recorded on a different planet. And that’s maybe one of the reasons why sound engineers who transferred the early 78 rpm records to vinyl from the 1960s to the 1980s had added artefacts so that they sounded more familiar to their comtemporary hearing habits. By the late 1960s it must have been almost 15 years ago since the last 78 rpm records were sold, by the beginning of the 1980s, these old records had already disappeared from the market 30 years ago and playing 78 rpm records had been by then impossible for the large public. Therefore a new media had to be used for public reeditions to be able to still listen to the Epoca de Oro tango repertoire.

This had been in the beginning exclusively the new vinyl record and later the tape and compact cassette. A lot of people I meet are unaware of this circumstance and very much focused on the CD editions as the ultimate reference. Often they try to show me some hidden logic in how the tangos are distributed over a particular CD. Whereas the CDs often represent just a lot of 78 rpm records put together in a chronological order sometimes covering all 30 years of the larger Golden Age but in an uncomplete manner as a tango CD typically holds only around 20 to 30 tracks. This batch is often put together on an availability basis and rarely in an editorial manner. There are some exceptions though like the English Harlequin CD collection. In reality the reference is neither the CD nor the vinyl or tape. The reference is the original 78 rpm record release for mostly every published tango before 1955. A 78 rpm record looks in size like a vinyl LP record but contains only one track per side. After 1955 started the vinyl era and some of the still present tango bands recorded then vinyl albums. In the case of D’Arienzo there were often a series of EP records and at a later stage a LP regrouping the EP releases. Di Sarli’s last recordings with Philips were later published on a single LP. I have a 1980’s reedition of this stereo record.

Grampian 636 Reverb Unit 1966

A typical artefact which had been introduced back then to the original recordings, is called a reverberation effect. While listening through the same titles of my collection I can distinguish versions with and without an echo effect. Famous examples are some of the early D’Arienzo recordings with Rodolfo Biagi on the piano. The initial recordings date from 1935 to 1939. If you listen to the echo containing versions, you have the impression that the orchestra must have been quite huge, recorded in a big hall with a lot of musicians. Whereas the standard version without the echo effect, sounds like very close and intimate. If I compare it to classical music, the echo version is the symphonic orchestra and the version without echo the chamber music.

The first known delay effects where available since around the 1950’s and were called Tape echo. To be more precise the echo effect is called a delay effect. With reverberation there are multiple delays and feedback so that individual echoes are blurred together, recreating the sound of an acoustic space. And I think this was precisely the intention of the sound engineers when they later added the echo to the old recordings: To add more depth to the flat sounding recording of the mono era. With the upcoming stereo culture of the 1960’s and 1970’s they wanted to give these old records a face lift and make them enter into the stereo age. As a side effect the echo also camouflaged imperfections of the original sound recording. And indeed when you listen to the echoed versions there is nearly no surface noise left. They make a clean impression.

Let’s have a closer look to the edition history of some of these tangos.

El cencerro performed by Juan D’Arienzo in 1937 is in it’s original 78 rpm record version of course without any echo effect. Later the same label, RCA Víctor, reissued El cencerro on a LP sampler called Evocando el Ayer, Vol.4, Serie Apoyando el Buen Tango CAL-3234. The 6 D’Arienzo instrumental LPs of this collection are very accurate reproductions of the original 78 rpm releases. They just contain some dating errors and some have no date indications at all but the sound quality is outstanding with no to very few reverberation effects and close to perfect tuning. This series had been issued somewhere in the mid to late 1970’s. In 1980 El cencerro had been again republished by RCA Víctor, this time on a 10 LP set called Serie Tango De Ayer, Vol. 3 D’Arienzo Biagi, containing a strong echo effect. It’s interesting to see how this tango is transfered later to CD: The El Bandoneón CD edition El Rey Del Compás EBCD43 (1999) as well as the Serie Tango Argentino CD Juan D’Arienzo Sus Primeros Exitós Vol. 2 (around 2004), seem to be based on the 1980 echo remix by RCA Víctor.  Also Milonga del corazón is the same version on the newer RCA Víctor BMG CD El Rey Del Compás 70 años El Esquinazo 1937-1938. It looks very much as if this LP had been the initial matrix for all follow-up D’Arienzo echo containing transfers. Therefore most of the dancers in America and also in Europe know only (or at least only had access to) the echo version as you can see on nearly all Youtube dancing videos like this memorable performance by Puppy Castello and Graciela Gonzalez from 1991 (I guess the music is played from a turntable or cassette deck and just from that aforementioned 10 LP set Serie Tango De Ayer, Vol. 3 D’Arienzo Biagi as neither the El Bandoneón nor the Tango Argentino CD existed at the time of this video!):

[youtube]http://www.youtube.com/watch?v=6f_BYsvVUz4[/youtube]

78 rpm master record

Please see also in this context the mystery about the Don Esteban recording as described on Johan’s website and which appears to be the only ever reedition of this tango by RCA Víctor. According to Johan it looks like the masters might have been lost. This is particularily sad as this reedition and all others have this strong echo effect and the El Bandoneón and RCA BMG CD copies inherrited it! Also this version of Don Esteban appears to be slightly different to the japanese CTA version which would suggest that there might have been two seperate recording sessions back in the 1930’s …

See here the sound samples:

• Don Esteban Juan D’Arienzo 1936 3:25 (with echo) From the LP Serie Tango De Ayer, Vol. 3 D’Arienzo Biagi

• El cencerrro Juan D’Arienzo 1937 2:41 (with echo) From the LP Serie Tango De Ayer, Vol. 3 D’Arienzo Biagi

• El cencerrro Juan D’Arienzo 1937 2:42 (without echo) 1:00 Extract of the Audio Park CD version

In a way the echoed version of El cencerro made it into the salón and became some kind of accepted sound icon.

cencerros

Maybe as a tribute to Juan D’Arienzo as he only made one recording of this particular tango back in 1937 when the recording techniques were still evolving. The echo version could be seen as an attempt to port the thin 1937 recording to a more space filling version as represented by the late D’Arienzo repertoire.

On the other hand nothing prevented the King of the Beat to publish a monumental version of El cencerro during his lifetime recording sessions which covered still another 20 years during the full frequency range stereo era from 1955 to 1975 — but as a matter of fact he didn’t, maybe he just forgot to do so … 🙂

If you want a version without the delay remix, you need to get Audio Park’s Epoca De Oro Vol. 1 APCD 6501 or CTA’s Juan D’Arienzo Vol.3 CTA-303. The japanese editions are based on their own 78 rpm transfers.

Now with this particular case presented in detail, I don’t know in how far people are aware that the echoed version is a pure invention? From the strict point of view of authenticity the echo version is untrue, some would say a lie. Back in the end of the 1970’s it had another meaning as we have seen and just until a couple of years ago it had been the only available version of this tango on second generation media, LPs, CCs and CDs! A lot of music in that time used extensively delay effects. Like on the 1977 David Bowie album Low which is interestingly from the same record label as the Juan D’Arienzo recordings, RCA Victor. This suggests that there must have been an interventionalist sound engineer at this company who added all sorts of artefacts to the historic recordings: Echo, tempo acceleration, low pass filters, etc. As to update these tunes to the taste of the day. And indeed other lables have been much more fidel when they restored and reedited their historic tango repertoire. In comparison EMI who inherited the ODEON catalogue had issued mostly very correct and authentic sounding LPs …

[youtube]Ihttp://www.youtube.com/watch?v=mkNmilE9ibk[/youtube]

Another added artefact to the original recordings is incorrect tempo as I just mentioned. This is actually by far a wider spread problem than echo effects, too strong noice reduction, wrong preamplification schemes or other record problems. The incorrect pitch which arises from incorrect record transcript sessions is the kind of problem which remains often unnoticed, but seems to haunt the listener unconsciously. Let’s see an example: Age Akkerman has described recently on his blog a case of an out of tune recording, Di Sarli’s interpretation of El ciruja. He  explains that he didn’t like this tango before and how retuning the tango to the correct pitch changed his mind and appreciation of this tango. Different tonal tunings can create different feelings or emotions in the listener. This can lead in the worst case to refuse and unlike a track. Most of these tonal shifts are due to a too fast running turntable during the transfer session. If you speed up or slow down a turntable you will hear how the keys of the music move around and change. Nobody really knows how these sometimes quite huge differences made it into the transfer recordings, it could have been via damaged transfer turntables or on “an early reel-to-reel tape recorder by changing the diameter of the capstan drive shaft, or using a different motor” [Wikipedia].  Nor do we really know if these changes had been done intentionally or by accident. But it seems that a lot of tangos are affected. Here is a quick test you can do in your own music collection to find such pitch problems: select the same tango, same recording date but from different sources/CD editions and check the duration and BPM fields. Are they different?

A couple of days ago I have recognised that there is something wrong with the pitch of Carlos Di Sarli’s Cosas olvidadas which sounded just too fast. Through an analysis of the partition, determinating the original key most proberly used by the band in their arrangement and carefully listening to the voice in the recording, while slowing down the track, I have pitched down the track by a total factor of around 2.5%. Now it sounds natural. For a comprehensive guide to retuning tracks, please see Age Akkerman’s paper Lost in Keys and check the examples on his website. This kind of pitch shift can get really dangerous when women voices are involved. They are already very high and if they are then pitched up further they arrive in regions where they can produce a pain in the ear (See my article on Too fast El Bandoneón recordings).

See here the different version of Cosas olvidadas as selected from my music collection for a comparison:

Cosas olvidadas Carlos Di Sarli : Roberto Rufino TA Canta Roberto Rufino Sus Primeros Exitos Vol. 1 Duration 2:19
Cosas olvidadas Carlos Di Sarli : Roberto Rufino RR Various Artists 1941-1956 Vol. 3 Duration 2:15
Cosas olvidadas Carlos Di Sarli : Roberto Rufino BATC-DIEGON Milonga Del Sentimiento Duration 2:21
Cosas olvidadas Carlos Di Sarli : Roberto Rufino El Bandoneón El Señor Del Tango Duration 2:17
Cosas olvidadas Carlos Di Sarli : Roberto Rufino Retuned version from Sus Primeros Exitos Vol. 1 Duration 2:23

Now one could say that these different speeds don’t have any effect on the audience but I have talked to some dancers and it seems that often these pitch shifts are noticed, in most of the cases unconsciously, they say then, the music is not so nice or restless. With some DJ collegues I also recognised that they won’t play repertoire of which they have a lot of too fast transfers. they say then: “Ah, Donato, I don’t like” or “Late D’Arienzo is too hysterical”. Other DJs like these faster versions if they need to mobilise energy in tired dancers, then again I think it would be an advantage to first correct the tunings and to operate in a second step a further up pitching while locking the keys, this can be done live with modern dj programs like Mixxx or Traktor. That way tangos can be speed up without affecting the tonal keys.

• Quintessentially find here a in D retuned version of El cencerro by Age Akkerman based on a track taken from the Remy Kooij collection (version with a light echo).