This week we bring you a second release of Rubinstein's Chopin, following on from the release a few weeks ago of his interpretations of the Piano Concertos.

The main aim here was to do something with the only recording he made of the complete 24 Preludes, Opus 28. These Rubinstein recorded in New York over the course of three days, each a week apart, in June 1946 for release on American RCA VIctor 78s - their British release came a full nine years later on HMV LP, earning high praise from The Gramophone: "this masterly performance is the best we have had so far of Chopin's great work".

Why Rubinstein didn't return to the studio with the Preludes, given their importance in the Chopin repertoire and the advent both of the vinyl LP, taped recordings and stereo, I do not know. But as it stands this is the Rubinstein recording; perhaps he felt his performance couldn't be bettered or even matched?

I decided to couple it with a 1959 recording, made in stereo using tape rather than direct-to-78rpm mono, and then see what might be done to lift the quality of the older recording up towards that of the later one.

That 1959 recording was indeed exemplary, from the golden age of RCA's Living Stereo era. If it had any flaws they were minor - slightly sterile acoustic, perhaps, leading to a tone where some body was lacking. Thus my remastering of the Ballades recording of 1959 has been, for me at least, rather minimalist! I did find improvements, but nothing too dramatic was needed here.

The Preludes were another matter. As I wrote a few weeks ago on the subject of Bruno Walter's Mahler, this is a set of recordings which required a much longer time-scale than normal, and a willingness to put them to one side and come back afresh a week or two later with new ideas as to how best to proceed.

The first part of the restoration process was pretty straightforward, with click and crackle removal software doing an excellent job. Although Rubinstein's piano sounded pretty even, Capstan pitch stabilisation software was able to iron out minor wow and flutter and really solidify the tone - one doesn't want the slightest hint of vibrato on a piano, after all!

Next we come to the equalisation and noise reduction part of the process. This is where problems within a source recording can suddenly have a sharp spotlight shone on them, with flaws previously buried in the poor tone of an older recording brought to the surface as the tonal shortcomings of the recording are undone.

In the case of the Preludes, some faired better than others. Indeed some sides were exceptionally clean and trouble-free. Elsewhere a little peak distortion was inclined to rear its head, and I tried various strategies to reduce or eliminate this altogether.

The ultimate aim was to raise the quality of the Preludes as much as possible towards the exceptionally high standards set by the Ballades, whilst acknowledging the limitations of the older recording. What I didn't want was a major lurch "back in time", or down in quality, as one set of recordings ending and the next began.

Fortunately the first Prelude was one of the best-preserved. Surface noise is minimal, piano tone is good, and one might be winding the clock back just 5 years, rather than to another era altogether in terms of sound quality. And so brilliant is Rubinstein's interpretation that by the time you do reach the odd note that's not as clean as you might like, several preludes into the work, you'll most probably find yourself so taken up by the performance that any sonic deficiencies will be of little distraction.

The piano tone lacks upper-end brilliance, as one would expect from a recording of this vintage, but otherwise it's pretty good, thanks to the XR re-equalisation, which provides a marked improvement over the rather thin and flat original.

What is especially fascinating for me is to compare the piano tone of the finished XR-remastered Rubinstein to that of the 1934 Cortot we issued late last year (PAKM059). Both have been re-equalised using the very same reference recording, yet the two couldn't sound more like two different pianos. Cortot's sounds more up-front, with the microphone capturing a much brighter sound as a result. Rubinstein's is a more mellow instrument placed further back from the listener.

What XR remastering achieved was to make both sound more like the real pianos they were - not to make them both sound like the same piano. And listening to the Rubinstein as I write this, he is utterly engrossing...
A bit of Science from today's newspaper...

 

Brain's music pleasure zone identified

• Alok Jha, science correspondent
• The Guardian, Thursday 11 April 2013 19.32 BST

Scientists know that music can give intense pleasure by delivering chemical rewards in the brain that are equal to the joy of good food or even sex, but now they think they may have identified the part of the brain where this pleasure starts.

Researchers scanned the brains of subjects while they listened to new songs and asked how much they would spend on buying the tracks. They found that the most popular songs - those which people were prepared to pay more for - were also the ones that elicited the strongest response in the nucleus accumbens, a structure in the centre of the brain that is involved in reward processing.

"This area is important because it's involved in forming expectations and these are expectations that could be rewarding," said Valorie Salimpoor of McGill University in Montreal, Canada. "What makes music so emotionally powerful is the creation of expectation. Activity in the nucleus accumbens normally would indicate that expectations are being met or surpassed."

In the experiment, which is published in Science, she and her colleagues scanned the brains of 20 people who used an iTunes-like interface to listen to 30-second clips of songs they had never heard before but were in a genre they generally liked. "Instead of just asking them if they liked the music or not, we gave them a chance to buy the music because that gives us a real understanding of what they really like and want," she said. "Immediately after they hear each clip, they make a decision. They could spend zero dollars, 99c, $1.29 or $2."

The brain scans showed a direct relationship between how strong a response someone had in their nucleus accumbens to a song and how much they were willing to pay for it. This part of the brain was not acting alone, however. Salimpoor also found that it was taking in information from the superior temporal gyrus.

"This part of the brain is the part that has stored all the templates of the music we've heard in the past and will be unique for each individuals," she said. "The way that we like music is 100% unique to who we are and what we've heard in the past and the way that our superior temporal gyrus has been shaped. The brain is working a bit like a music-recommendation system."

The latest results shed further light into Salimpoor's 2011 study, which found that the experience of pleasure when listening to music was mediated by the release of the brain's reward chemical, dopamine. She said that music seemed to tap into the circuitry in the brain that had evolved to drive human motivation. This ancient reward system, when listening to music, was being used to provide a cognitive reward.

Prof Sophie Scott, a neuroscientist at University College London, cautioned that Salimpoor's results should not be over-interpreted. "It is clearly the case that you get rewards for the music you like [but] I don't think we listen to music in any one way, we listen to music in the same way we read books or read poetry or engage with other sorts of art," she said. "One of the reasons they are things we like is because we can engage with them in multiple ways - you could be enjoying music because of the rhythm, because of the way the singer's singing, there's so much going on.

Reward was only a snapshot of one particular brain system and its involvement in music, Scott said. "But don't think it's telling you everything about the totality of how your brain engages with music."