In the previous chapter we noted that the transfer quality of the same song can vary widely. This depends on many factors, the quality of the original shellac records, the technique used for recording the music and transfering it to digital format, and the post-production where equalisation can be used to counteract the lack in frequency responce in the media involved and to clean up the sound. We can assume that all of these techniques are used by different producers but in varying measures. I feel that it is important to research these effects in order to have a clearer understanding of how to adjust our music so that we can present it in its utmost variability, balanced but detailed at the seme time. Here I am using frequency graphs made from 4 songs in order to see how different versions have been affected in both transfer and through re-mastering. The songs are (1) Queja indiana by Orquesta Francisco Canaro from 1927, (2) Cantando by Orquesta Adolfo Carabelli from 1931, (3) Grand Hotel Victoria by Orquesta Juan D’Arienzo from 1935, and (4) Recuerdo by Orquesta Osvaldo Pugliese from 1944. In all cases will be using the TangoTunes version for comparison. The reason for this is that they provide very distinct information about their transfer process, a state of the art transfer, where they adjust for the inherent changes in frequency response made during the production of shellac reords, but do not re-master the tracks further, except by cleaning surface clicks. I therefore take these tracks to be as close as one can get to a “true” transfer, and therefore suitable as reference tracks.
1. Queja indiana by Orquesta Francisco Canaro from 1927
This is an early electronic recording made by Francisco Canaro in 1927. I will be comparing the TangoTunes version with three others in an attempt to discern the differences in transfer technique and mastering.
This is a spectrogram, via Infinite averaging, of the first 60 seconds of Queja indiana. We and see that the most intense volume is from around 180 to 700 hz, with the bass drop-pping off gradually to around 40 hz and the midrange being reasonably balanced from around 700 to 3000 hx, with limited variation from 4500 hz onwards. It is also noticable that the low-mid frequencies (200–500 Hz) are the strongest, with the bass (80–200 Hz) and high-mid ranges as much as 20 dB lower (note that tango music at this time does not extend upward into the treble range, from 7–20 kHz).
This is a comparison with a version from Tangos – Francisco Canaro – EPM 995322. We can see that the low frequencies have been shelved from around 50 hz down, and everything above 4500 hz has been shelved as well. This is the range that is outside the frequency responce of pick-ups before 1930. We can also see that the bass ranging from 50 to 190 hz has been intensified by up to 3 db and the mid and hig-mid ranges, from around 1000 hz up to 4000 hz have been amplified by up to 7 db. To sound similar to the TangoTunes version both the bass ranges and the high-mid would need to be reduced quite a bit.
This a track from an LP called Memorias del Trio Argentino – Memorias LP585, an example of a typical low-budget production. Here a lot of the bass information has been cut off, from 90 hz downwards. the same applies to the high frequencies, cut off at 3 kHz to reduce hissing. Here we also see that the higher frequencies have not been amplified in the production, being up to 18 db lower that the TangoTunes version. This is probably because the stylus used for the transfer was not suitable for the shellac original and therefore had limited sensitivity to the high frequencies. Due to lack of information this version could not be salvaged by equalisation.
This is a version from the Noche De Reyes – 4º Concurso De Tangos De Max Glücksmann Y Sus Discos Nacional 1927 – El Bandoneón (EBCD-152). Here the quality is much better than in the previous version, although the premises for post production are interesting in comparison with the TangoTunes version. We see that neither the bass has not been shelved. The bass is less intense, it is about 4–5 dB below that of TangoTunes. The high-mids are shelved from 2 kHz onwards, thus losing importantdetail between 2 and 4 kHz. To counteract this the range from 700 to 2000 Hz has been intensified by around 5 dB. For this to sound similar to the TangoTunes version the bass would need to be intensified by about 5 dB and the mids reduced by 5 dB, with a increase of 5 dB from 2300 to 3300 Hz.
2. Cantando by Orquesta Adolfo Carabelli with Alberto Gómez and Mercedes Simone from 1931
This is a song from 1931 that has been trensferred by many different producers. First I will compare its frequency curve with the previous example, an indication of improvements in recording technique from 1927. Then I will compare the different prevailent versions of this track to the TangoTunes version.
Here I’m comparing Canaro’s 1927 Queja indiana (blue) to the TangoTunes edition of Adolfo Carabelli’s Cantando from 1931 (green). What is evident is that the effective range of these two recordings seems to be similar, with usable sound ranging from around 70 Hz to 4 kHz. It is noticable, however, that the 1931 recording is more balanced as the bass and high-mid registries do not fall off as much as in the 1927 recording.
Her we compare the TangoTunes version of Cantando to the version on Adolfo Carabelli Vol. 2 – 1931-1932 – (TDJ-1019) from Tango records / Tango-dj.at. Here we notice that the TDJ version does not increase the bass level as much as TangoTunes does, being around 6 dB lower around 100 Hz. The higher frequencies, from 2–4 kHz are more enhanced, by roughly 4 dB. While TangoTunes leaves the upper and lower registries unchanged, TDJ shelves these below 70 Hz and above 4 kHz.
This is a comparison with the version on Adolfo Carabelli 1931-1934 – BATC (ORQ 247) from the Buenos Aires Tango Club (the same as the one on Memorial Del Tango 6 – Adolfo Carabelli – AMP (CD-1162M) from Yoshihiro Oiwa) . In this version the bass seems not to have received any adjustment after transfer, being about 10 dB lower at 100 Hz. The treble drops off a bit after 2 kHz, sloping down from there losing some detail.
This is a comparison with the version from Adolfo Carabelli Vol. 1 (1931-1933) – (CTA-261), Akihito Baba’s Club Tango Argentino version. This is close to the TangoTunes one. Here however the bass has been raised a little bit less, by about 4 dB, while the high-mid has been more attuned, by about 3 dB. Bhabha also shelves the higher frequencies above around 3.5 kHz.
This is the version from Adolfo Carabelli – Vol 1 – 1931-1932 – Sello RCA Victor – Club De Tango. This version seems unbalanced compared to the others. I would suggest that this is due to an inferior transfer technique, with the stylus not able to follow the grooves as closely. The range above 300 Hz is a bit dull, while the bass is, as in most of the other trensfers, lower than in the TangoTunes version. The high-mids also start to drop a bit early, from around 2.5 kHz, leading to loss of detail.
The final version is from Adolfo Carabelli y su Orquesta Típica 1931-1933 – Cuatro Palabras – El Bandoneón (EBCD-87). As we can see it is very close to the TangoTunes version above 200 Hz. The bass however is a lot lower, by around 9 dB at 100 Hz. The high-mids are also shelved a little bit early, at 3 kHz, losing some detail.
We can see that most of the transfers available for Carabelli’s Cantando seem to be of similar quality. The variations are small in the upper registries, but more observible in tha bass frequencies. Here most producers seem to be following a different adjustment curve from TangoTunes, not raising the bass level nearly as much as they do. If this is indicative of transfers for this period it would suggest that the DJ would need in general to raise the bass level of songs from this period by about 6–9 dB.
3. Hotel Victoria by Orquesta Juan D’Arienzo from 1935
This is an early tune from D’Arienzo’s main orchestra, from 1935, the beginning of the Edad de oro. Again I use the TangoTunes version for reference, comparing 5 other versions to it to demonstrate the different approaches of transfer and production by different producers.
This is a comparison between the TangoTunes versions of Cantando by Carabelli from 1931 (blue) and Hotel Victoria by D’Arienzo from 1935. The first one notices is the similarity between the detail in the two curves. This would suggest that there have been no great differences in recording quality between 1931 and 1935. The balance is similar and the detail suggests a practical range from 70 Hz to 4 kHz, as before.
Here is a comparison between the TangoTunes version (in blue) and a version from Juan D’Arienzo Vol. 01 (1935-1936) – (CTA-301) (green) by Akihito Baba. They are remarkably similar, especially in the middle range. TangoTunes has a bit more intense high-mids, about 3 dB higher from 600 Hz up to 3000 Hz. In comparision the CTA has a bit (1.5–2 dB) more intense bass, from 200 Hz downwards. The CTA version, however, cuts the high frequenzies from 3 kHz upwards, probably losing important detail in the range from 3–4 kHz.
Here we have Hotel Victoria from Juan D’Arienzo Época de oro Vol. 01 (1935-9138) – (APCD-6501) from Audio Park in Japan. Here the high ranges become gradually lower from 500 Hz upwards, being 5 dB lower around 2500 Hz, the frequencies above that being radically lowered. The bass is also severely reduced from 100 Hz downwards.
Here we have yet another quality transfer, from Siglo Del Tango Argentino CD 04 – Juan D’Arienzo – (BVCP-8704) by BMG Victor Japan. We can see that the differences to the TangoTunes version are very subtle, the high frequencies a trifle lower, by between 2 and 2.5 dB upwards from 2300 Hz, and the bass a bit more intense, about 3 dB higher below 200 Hz. The cutoff after 4 kHz is also a bit more rapid than in the other versions.
Finally we have an edition from El Bandoneón – Blue Moon, Las Grandes Orquestas Del Tango / 40 Grandes Éxitos – Juan D’Arienzo CD 1 – Blue Moon (BMT 610). Here we have similarities to the CTA-version, with a stronger bass and weaker high-mids. Of these 5 versions the higher frequncies are the weakest here, being around 2 dB weaker at 1000 Hz and being about 5 dB weaker at 4000 Hz. In order to match the TangoTunes version here we would need to strengthen the high-mids gradually and lighten the bass a trifle.
Comparing the different versions of Hotel Victoria one notices that they tend to be quite similar in scope. It is however noticable that many tend to make the higher frequencies a bit weaker than the reference copy, indicating that these would be a bit “duller” in the upper ranges. A slight strengthening from 2–4 kHz would make these copies more similar to the TangoTunes version.
4. Recuerdo by Orquesta Osvaldo Pugliese from 1944
This is a tune from the mid-forties, when recording technology had improved quite a bit. This is also one of Pugliese’s greatest hits and available in many versions. First we will compare the TangoTunes version of this tune with D’Arienzo’s Gran Hotel Victoria from 1935, to see how the shape of music has changed.
It is interesting to see how TangoTunes’ Recuerdo from 1944 compares with Gran Hotel Victoria recorded 9 years earlier. We notice that in general the sensitivity across the spectrum is similar. It is however evident in these examples that the range of the spectrum has increased. There is more detail in higher frequencies; in 1935 there is usable detail up to 4 kHz while in 1944 this extends up to the treble range, to 7.5 kHz; in the bass we seem to have more detail down to around 40 Hz compared to 50 Hz in 1935.
This is the version from Colección – Osvaldo Pugliese – EMI CD 796624. It is incredible to see how different this version is. In general it seems to be a lot more impalanced. Compared to the TangoTunes version the bass is much lower; it seems to have been cut between 4 and 12 dB below 150 Hz. The range from 1 to 5 kHz is also a lot lower, between 4 and 9 dB, with a strange peak around 7 kHz, as if it had been specifically enhanced around that frequency.
This is from Grandes del Tango 02 – Osvaldo Pugliese CD 1 – Lantower 10015. It is similar to the one before, except that the lowerring of the high frequencies is more gradual, with the peak at 7 kHz not so sharp.
This version is from Instrumentales Inolvidables – Osvaldo Pugliese – Reliquias (859023). In this version it seems like the higher frequencies have not been enhanced but gradually lowered from 1.7 kHz onwards. It is obvious thie version runs faster than the TangoTunes one, i.e. the higher frequencies are markedly higher. The frequencies above 10 kHz are also abrubptly shorn.
This is from Los Clásicos Argentinos – Tango – Volumen 04 – Osvaldo Pugliese: bien milonga. It is more balanced than most. Although there is a drop in volume around 2 kHz the higher frequencies stay stable until the 7 kHz mark. after that the cut-off is quite sharp. It is difficult to speculate on the reasons for the way these four different versions of this tune are all a bit different but still with similar main differences from a more stable version like the TangoTunes one. I think it would be plausible that one of these, like the EMI one, was close to the original transfer and that all the others have been remastered so as to reduce the obvoius faults in that one, with very disparate results.
This version, however, is different from the rest and clearly more closely similar to the TangoTunes version. This is from Osvaldo Pugliese Vol. 1 (1943-1944) – (CTA-521), produced in Japan by Akihito Baba. Here most of the frequencies are much more balanced and in a similar tempo as TangoTunes. There are slight differences in tha bass being a bit more pronounced in this version, while the higher frequencies are about 1.5 dB lower up to around 3.5 kHz, when they fall of a lot more. It seems like Bhabha starts shelving the upper frequencies at 3.5 kHz, as if this were a standard procedure all across the spectrum, disregarding the increased sensitivity of recordings after 1940.
In examining the different versions of Recuerdo we become aware of a characteristic applying to many commercial productions of popular tango tunes. While the “audiophile” transfers of TangoTunes and CTA try to be “true” to the original, the commercial versions seem to follow a different logic. One can speculate that these all derive from the same master, perhaps a magnetic tape master made for transfer to a vinyl record in the late fifties or early sixties. This would have beenmade available to the different producers, they attempting to re-master it to correct for the evident faults originating in the original transfer. Thus we have a number of different copies that very probably derive from the same faulty source. If a DJ has to rely on one of those, he will have to be very careful in equalising in order to get a more balanced version. His best option, though, would be to try to get a better copy to begin with.
So, our study has led us to conclude a few things about the differences in balance and scope of various transfers of tango music.
In general quality transfers tend to correct for lack of bass or certain higher registries in transfer from shellac records. It is therefore not any reason for DJs to correct against these in performance.
The scope of earlier recordings, made before 1931, is more limited than that of later ones. This means that the bass and high-mid registries are weaker than the central ones. This may in some cases mean that it would sound better to accentuate these in performance.
A number of transfers either under or over-represent the lower and higher registries. This would mean that in these cases the bass or high-mid should, as the case may be, either be amplified or reduced discretely to attain better balance. This is dependent upon the quality of each track.
It is not uncommon for even quality transfers from more recent recordings, to reduce and slope the higher frequencies, probably to reduce hiss, thus losing detail from 3500 Hz and above. This may mean one needs to compensate for this in performance.
Transfers of popular tracks tend to be reproduced in many different versions where different techniques of remastering have produced uneven and often unexpected results. This may be difficult to counteract, although careful equalisation may alleviate some of the worst problems. In these cases, however, it would be preferable to try to obtain a better balanced and more direct transfer.