Research into tango has to a great degree focused on the the aesthetic and social functions of music and dance, both individually and culturally. Little focus has been put on the actual technical underpinnings of the material we have to our disposal; the historical recordings that are played at countless milongas all over the world. There is therefore a need to gain an understanding of key issues in terms of the production of the original records and how these issues influence their reproduction and release in the present.

One matter has repeadedly been discussed in technically oriented tango circles recently: how recording conditions and practices may have affected the sound of the original shellac recordings. These considerations include a discussion of the dynamic range of the original recordings, and how that evolved. That issue I have dealt with elsewhere, although it will need to be addressed more thoroughly later on. Another pressing matter concerns the tuning of the orchestras, on finding the ‘optimal’ pitch to present the music as faithfully as possible. In this regard there are a lot of conflicting opinions circulating, especially since factual evidence is regretfully lacking. The present research is thought of as a starting point in addressing these issues.

Premises

What is involved here is an analysis of the pitch present in a distinct set of tango shellac recordings, dating from 1926 until 1950. The material at hand is limited, comprising the Argentine shellac recordings in the Internet Archive, digitized by George Blood. As these recordings were digitized using precise equipment we can be assured that the speed for the digtisation process was excactly 78 rpm without any corrections being made afterwards. The analysis of tuning pitch was done in Sonic Visual from the University of London, using the softwares tuning analysis module.

The dataset involves two major parts, 108 recordings by the Victor label and 121 by the Odeon label. The Victor series includes recordings by Orquesta Típica Victor, Carabelli, D’Agostino, Ferrazzano, Troilo, Di Sarli, Lomuto, and D’Arienzo. The Odeon series includes recordings by De Angelis, Francisco Canaro, De Caro, Caló, Fresedo, Pugliese, and Firpo.

The Victor dataset

The pitch of the Victor dataset, from 1926 to 1951, 108 recordings in all.

Viewing the entire Victor dataset it is obvious that something has changed around 1939, as the pitch becomes more uniform and variations, although still present, decrease. From 1939 onwards the median pitch also rises a little.

The Victor recordings from 1926-1939, before the apparent change in pitch variation.

In the period from 1926 to 1939 we see that the pitch variations are greater than later on, although the Victor pitch is hveretheless quite uniform, wit most recordings being in the range from 435 hz to 440 hz. The trendline shows that there the mean pitch over the whole peeriod is rougly the same, at around 438 hz. A probable cause for the variability in pitch over this period is the inherent variability in the electric mains current at the time, which is known to have caused problems in calibration the speed of the master disc recording lathes. As we see in the next image it is porbable that Victor introduced new equipment in 1939 that could have made the calibration of the lathes much more reliable.

The later part of the Victor dataset, from 1939-1951, after the apparent change in variation.

Here we see the increased stability of the Victor pitch after 1939. A lot of the samples are pitched close to the average of 442hz. It is interesting to note that while variations in pitch are very small in the beginning of the period they do increase in 1946. One wonders whether ageing intruments were proving more difficult to calibrate or whether the orchestras themselves were increasing the pitch they tuned to.

The Odeon dataset

The pitch of the entire Odeon dataset, from 1926–1951, 121 recordings in all.

Unlike Victor, the Odeon dataset shows a lot of changes in mean pitch over the period involved. There are two important changes that are obvious. In 1927 and 1928 there is a lot of variability in the pitch, but the median pitch is still around 440 hz. From 1929 to 1944 the variability continues, but with a much lower median pitch. After mid-year 1944 there is a great change yet again.

The Odeon pitch from 1944–1951, after the change in pitch variation.

As we can see here the pitch for Odeon recordings from 1944 onwards is really stable around the standard of 440 hz, with surprisingly little variation. One would surmise that Odeon installed new greatly improved equipment in 1944 and adopted the standard of 440 hz with a rotation standard of 78 rpm at the same time.

The Odeon pitch from 1929–1944, displaying the lower median of around 430hz.

It is surprising to see the change in pitch for Odeon after 1929, when the median pitch becomes more than a third of a semitone lower than before, down to 430 hz. Research has shown that on the average the actual pitch during the first half of the twentieth century was thrroughout around 440 hz, just as the Victor recordings demostrate. It is therefore surprising if Odeon chose to adopt, in 1929, a much lower standard pitch when they, presumably, installed new improved equipment. The analysis, however, demonstrates a lot of variations in pitch from 1929 to 1944, ranging from 423 hz to around 440 hz.

The Odeon pitch from 1929–1944 shifted upwards by 2.56%, the difference between 80rpm and 78 rpm.

It has been mentioned in documentation about the actual practices of the Odeon company that they did attempt to adopt a rotation standard of 80rpm in the late twenties. One could, based on that and the evidence from their recordings, speculate upon whether they did so in Argentina in 1928 when installing new equipment and kept up the practice until 1944 when the mastering equipment was renewed. In the graph above and in the one below the pitch has been recalculated on these premises, that the Odeon recordings from this period were cut to be played at 80rpm. By doing this the recording pitch, although still showing a lot of variability, tends to be more uniform in the whole period in question, averaging at just over 440 hz. Of course this is just speculation, one that would need to be ascertained by going through documents relating to the industrial practice of the Odeon company in Argentina at the time.

The entire Odeon dataset, with the period from 1929 to 1943 shifted upwards.

Conclusion

So, to sum things up: When comparing the playback pitch of tango recordings from Victor and Odeon in Argentina we are faced with a dilemma relating to the very different profiles of the two companies. The pitch from Victor, although showing a lot of variations prior to 1939, is apparently averaging around 438–440hz throughout the entire period. The pitch from Odeon is in tune with Victor in 1927–1928 and from 1944 onwards, averaging at just above 440hz, the median pitch of around 430hz from 1929 until 1943 is however a dilemma. One can speculate about the reasons for this but it is certain we cannot find a conclusive reason for this disparity without further research.