Svetlanov re-recorded considerable amounts of his performing repertoire toward the end of his career partly as a memorial to his career but also from the conviction that only Russians could really play Russian music. However contentious that may be there can be no doubt that his recording of the 3 Rachmaninov symphonies and other non-concerto works forms a considerable artistic and musical statement.
This disc is a good example of the approach. Overall timings for movements are frequently on the longer side but this hides the considerable fluctuation of tempi within each movement. As a result the performances are more temperamental than many others with a heightened emotional impact. Excitement is frequently achieved through extra weight rather than extra speed. The balance of the works therefore becomes at variance to that with which we are often used. These interpretive characteristics apply equally to the sections of the Isle of the Dead.
The tonal characteristics of the orchestra are also different. The woodwind and brass have more of a 'raw' edge to them than is found in Western Europe or the USA fer instance. These characteristics are still to be heard in recent recordings of the Bolshoi or Mariinsky orchestras. This tonal range is far more tamed than in recording from earlier generations so this is more of a compromise but nevertheless it still sounds Russian by comparison.
The recording by Warner is very good indeed without any of the roughness of earlier Russian recordings or their unlikely balances. Everything is now realistic and can be counted as good recorded sound.
For anyone interested in Svetlanov's view of a Russian approach to this music I would suggest that there is much food for thought here. The performances are very convincing and offer a good alternative view. Whether it is actually more authentic or not I would not know. However I would expect most purchasers interested in this approach to find this whole series of Rachmaninov, including this particular disc, to be very well worth purchasing.