Gustav Mahler - The Symphonies

Gustav Mahler once said: "The further the music develops, the more complex the apparatus used by the composer to express his thoughts becomes." ...and the more difficult it becomes to record it, one is inclined to add. Mahler's works are indeed among the most demanding for the orchestra as well as for the recording and reproduction equipment.

Gustav Mahler
Symphony No. 5
Michael Tilson Thomas
San Francisco Symphony
96kHz/24bit download

Perhaps the most demanding task for me was actually finding the record that satisfies my needs as an audiophile. I owned several records of different Mahler symphonies from Rattle, Haitink, Tennsted and others and they were okay until I got used to the quality of Denon's Bruckner recordings. So, I then sat in front of my new loudspeakers and I couldn’t even get through the first movement of the version by Rattle and the Berlin Philharmonic. I got stuck between "okay, that sounds nice" and "why isn't this more transparent, is this distorted somehow?". Those thoughts bothered me long before I even began criticising tempi or other aspects of the interpretation. Thus I decided to undertake a big project with symphony no. 5 and go through many of the available recordings until I got to the perfect one.

One part of the second movement stood out for me as an especially difficult one in regard of sound clarity. It occurs around the 10th minute of the second movement, not even a moment of very high volume, but subtlety within the string section which becomes very distorted on most records. I purchased a bunch of 10 individual tracks of the second movement. Some sounded slightly better than others but none were really convincing. Furthermore, the listening experience was not the same from the excerpts I took at around the 10th minute (that’s the one you hear from the samples) as it was from the beginning of the movement, as if the sound quality differed within the piece for some reason.

At least that was my first impression, Abbado/Chicago sounded the clearest of the bunch. But further in and with longer listening, the sound appeared rather artificially sharpened and too harsh, maybe it is a remastering that went too far. Vänskä/Minnesota became my next best choice, a very recent recording as it seems. Sharp but less harsh then Abbado/Chicago, with more transparency than most other recordings, but I still waited for the orchestra to become real.

The records I compared (all from the 2nd. movement, around minute 10):

Uncompressed WAV in original kHz and bit-depth of the records, peak normalized *

So, I went and got myself 6 additional tracks to listen to and none of these really convinced me. I was about to give up when I decided to give Tilson Thomas with the San Francisco Symphony another chance. I had the 7th and 4th already but had never really listened to them because they had clicking sounds every couple of seconds. At first I thought the audio files from HDtracks had got corrupted but then I found that either the player or the DAC caused the problem out of that specific 88.2khz sample rate. Luckily, I eventually found the same records on Prestoclassical with 96khz and those really knocked my socks off – that sound was so natural and present, it spread out before me in the air with its seemingly unlimited potential of dynamics and detail. What surprises me the most (and like before with Denon) is that the perfect recording does again not come from one of the well-known, big recording labels, but from a widely unknown source, in this case it got produced by the orchestra itself on its own label SFS Media. We also see an example here that a high resolution can not make a record better, but a really good recording may benefit.

On this occasion I had the opportunity to compare the same record from different sources, one I bought from HDtracks where the best resolution was 88.2khz/24bit and one from Prestoclassical with 96khz/24bit. I analyzed both versions in my audio program and found out that the audio from HDtracks had a generally low gain level, less then half of that from Prestoclassical. Nevertheless, in some of the movements there appears at least one peak level that goes to the maximum, like at the very end with a heavy big drum or crash cymbals. With more then 200% gain on the 96khz files, they must have cut the extremes. But overall, the higher gain is favorable, because for listening the lower gain files, I had to turn up my amp to 70%, while usually I have it around 25% to 50%. The question arouse if one of the versions would be inferior in quality, so I took a sample, made peak normalization on both, compared the waveforms and saw they matched perfectly.

Well, it was a time-consuming exercise whilst I was already actually sitting on the very thing I was looking for. But, just in case you are wondering which are the better ones out of those 17: A long way after the top choice Thomas/San Francisco would be Wit/Polish, Chung/Seoul or probably Vänskä/Minnesota.

As for the other symphonies, I am glad this case is so clear and Michael Tilson Thomas did indeed record all the Mahler symphonies in this series.

*Audio © notice: Copyright of the audio excerpts from the respective records lie with the individual record labels. UK, EU and CH copyright laws allow short excerpts or citation as fair dealing with a copyrighted work for the purpose of criticism or review.