The media we have available, the digital files with tango music, form the basis for a quality DJ performance. There are things you can do later on, in equalisation, to improve a bad file, but that can only tackle smaller problems. It is therefore important to make sure you are working with the best available material. In that respect it is important to work with relatively high-quality files in general, to know where they came from initially, and to choose the best among many possibilites.
File quality
If you want to be safe in terms of the quality of your files, your safest bet would be on cd quality or lossless files. These are usually better than compressed files, such as mp3 files. That said, quality mp3 files 256 or 320 kbps can be of quite good quality since the compression involved is quite limited. These files can be of good quality, especially if they are made with good software directly from cd-s or from lossless files. In that case a compression to 256 kbps or more results in very little loss of quality. A greater compression, down to 128 kbps or lower, usually leads to quality loss, especially for older tango music. The reason for this is that compression algorithms are usually optimised a greater frequency range than older recordings encompass. They therefore tend to compress and simplify the frequencies important for tango music leading to a more “muddy” sound. Since music is usually not sold at lower compression than 256 kbps it is also likely that the provenance of such a file is doubtful. Such a file may be made directly from a cd, which would result in reasonable quality. It is more likely, however, that it is a recompression of an already compressed file. When a compressed mp3 file is re-compressed the distortion effects of compression are magnified, often leading to strange sounding “artefacts” in the music.
Transfer quality
Even though file quality is important, the source the file originated from is even more important. A quality file you can buy on a cd can be very different from another copy of the same performance on another cd. What matter here is in what condition the material was when it was transfered to the cd. Here the possibilities are numerous, depending on (1) how the transfer was made from the original record and (2) what sort of post-production (re-mastering) was performed after the transfer was made. Pre-1950 recordings, and a number of recordings after that, were made on shellac records and intended to sustain a lot of wear due to playing with heavy conical needles. A number of these recordings were transfered to other media in the fifties and sixties, copied from good quality shellac records onto magnetic tape, cleaned up and transfered to vinyl records. These magnetic tapes, or even the vinyl records themselves, are often the source material for more recent cd or digital copies. Today, the best transfers are still being made directly from collection shellac records, which are converted directly to digital media directly.
(1)
Older transfers may actually be quite good, especially if the process was well handled out. The available material was at the time quite recent; in the fifties one can assume that more pristine records were available for transferring the material so that would have led to better quality. We know, however, that the transfer techniques at the time were not really advanced. Most often the transfer would have been made using modern styluses that would not fit well to the wider grooves of the shellac records. The result was that not all of the sound spectrum was properly reproduced. We may also assume that transfer of older material was made at 78 rpm, which became standard later for shellac records. Many of these were actually recorded at other speeds and therefore would be higher or lower in pitch than the original. So, even though many of the older transfers are quite well done, technical knowledge at the time was less than now and the quality therefore very varied.
More recent transfers, from as early as 1970, but most made in the last 20 years or so, are made according to better technical specifications. The best transfers are now made with equipment specially made for transfers, with styluses specially made to fit the shellac grooves and with better analysis of the original speed of the records involved. Of course it is likely that fewer pristine records are available, sometimes the only copies of a tune are available on worn shellac records. With high quality masters contemporary methods will generate better quality material, in general, than the older methods were able to.
(2)
The problems facing those that produce digital media is complicated. In the era of shellac recording recording techniques were not standardised. As a rule however, bass frequencies were reduced during recording and boosted in playback; this way the grooves could be made narrower, allowing more time on the record. Limitations of microphone technology led to higher ranges being flat. Shellac records also include a lot of hiss in the higher ranges, where there is little musical material, and hum in the lower. The problem for DJing is that there is in transferring the material no standard way to deal with those problems. Very often the lower and higher ranges have been cleaned, to reduce hiss and hum, to varying degree. Sometimes the bass level has been boosted and sometimes the higher ranges have been made more prominent. In a number of “re-mastered” tracks reverb has been added, to create a more modern resonance in the recording and hide problems in quality. What this means for the DJ is that the same track can sound very different on different cds, even when the source transfer is the same. Some tracks are proficiently post-produced and sound quite good. More often tracks tend to be flat or to miss a lot of the higher and lower tonal ranges, when anti-hum and anti-hiss filters have been over-applied.
It is for this reason that many tend to gravitate to contemporary, more direct transfers. These are made with appropriate equipment and usually involve minimal post-production. They include the lower tonal ranges with their hum, and the higher ones with their obvious hissing sounds. This means that these tracks actually do not sound very nice if played without adjustments, as the hiss and hum make them difficult to listen to. With appropriate equalisation they can however be made to sound great, since they have a lot of detail and the difference between instruments and tonalities is more obvious than in more produced versions.
Examples
I’ve selected a few versions of the same song, Rodríguez Peña, from 1938 with the Juan D’Arienzo orchestra in order to compare different transfers and compressions. These are 12 seconds from the climax of the tune from different transfers where the difference in quality can be clearly heard. (Please note that these have been compressed to 256 kbps mpegs for this web. Since the compression has when possible been made from lossless files, there is not a discernable loss in sound quality.)
The first one is from the Audio Park publication, from the collection of Shigeru Terada, from a cd published in 2007. Tanaka was really precise in his technique and his D’Arienzo collection is generally of great quality. It is an accurate transfer with a good deal of variation in intensity and differentiation between individual instruments. The copy has been processed so as to reduce the crackle, but not so much as to effect the upper registries.
The next version is a more recent transfer made by TangoTunes, a producer in Vienna and Buenos Aires. They have been working with state of the art techniques of transfer since 2013. As you can hear it is as clear as the Audio Park version, but with a bit more sparkle and more hiss. TangoTunes do only minimal post-processing on their tracks and their tracks therefore will sound too harsh without equalisation. You can see from the curve that this version, as the Audio Park, is nicely balanced and varied.
This version is from the Album Ataniche from Euro Records produced by the Buenos Aires Tango Club in Buenos Aires, available for download on sites like Google Music. It is a reasonable version, especially if you are not able to adjust your equalisation yourself. It has been through a lot more production than the above two. The hiss of the record has been reduced and is barely noticeable except in the bright parts. The record has also been normalised, the soundlevel being adjusted to a maximum. The production leads to a slight loss of detail and there is less information available to accentuate with proper equalisation.
Here, however, we have a bad copy, a version available on YouTube. It has been over-cleaned and normalised way too much. It sounds as some clipping of volume has occurred sometime during the process and that too much compression has led to severe lack of detail, it sounding flat and uninteresting. One would speculate on the quality of the original transfer, whether detail was already lost there.
The final example is from a YouTube file as well. This, however, seems to be from a better source. It has however been equalised, the bass is a bit boomy, and there is some loss of detail. This however, unlike the one above, could be used in a milonga if nothing better were available.