From the back of the cover.
"This sonata's subtitle, 'Moonlight', was not Beethoven's but was adopted after the German critic Heinrich Rellstab had likened the first movement to moonlight on a lake. His description clung, and is now perhaps the most famous of all musical nicknames. The work was composed in 1801.
The slow movement begins the sonata. It opens with triplet arpeggios which persist throughout to the end, and above them a long, sustained melody soon emerges. There are occasional points of imitation between this theme and the bass of the piano, which echoes the rhythm in a sombre and menacing way. The movement is given almost entirely to this cantabile them.
The second movement is, in effect, a scherzo, for it moves considerably faster than the normal minuet tempo, but it is not as yet as vigorous as some of Beethoven's later scherzi. Indeed the movement, despite its speed, is rather graceful and charming and seems to betray its dance origin. Notice how the opening bars are immediately repeated with syncopations and layered phrasing. Unexpectedly the trio section begins more violently than the Allegretto itself and is very rapid in its changes of mood. After the trio comes the usual repeat of the first section.
The dramatic, urgent third movement is in fairly straightforward sonata-form. The first subject, announced immediately, consists of groups of rapid ascending arpeggios, each culminating in two heavy sforzando chords. After a pause the quieter, more legato second theme is introduced, but it is quickly displaced by rushing scale passages. Later still a further legato phrase is heard, which soon gives way to a return of the opening arpeggios. There is a good deal of development of the first two subjects, and eventually the second of the legato phrases is reintroduced, after which two pauses hold up the flow of the music. With elaborate arpeggios and a chromatic scale, the tempo gradually slackens and the music sinks to rest for a few moments, only to lead to a short, vigorous close."
|Composer||Ludwig van Beethoven|
|Work||Piano Concerto No. 14 in C sharp Minor 'Moonlight', Op. 27|
|Date Recorded||Circa 1958|
|Date Transferred||29th of April 2012|
|Date Restored||3rd of March 2017|
|Serial Numbers||Decca CEP561|
|Bandwidth||45Hz to 9kHz|
|Transfer Stylus||0.7 mil elliptical|
|Transfer Cartridge||Ortofon OM5E at 1.8g|
|Transfer Turntable||Project Debut 2 at 45RPM|
|Click Reduction||DeClick, wavelet mode, 60, 2 passes|
|Limiting Filter||48dB/octave Butterworth high pass at 45Hz|
|Additional Eq.||First order roll-off at 8kHz|
I had originally transferred this disc to the digital domain in April 2012, but it sat in my to-do folder for almost 5 years, most likely due to the rather poor sound quality for its age. There are quite a few very obvious defects that you can hear throughout the recording. Distortion is higher than I would have liked it to be, but this disc has probably been played many times with a heavy and possibly worn low compliance pickup. There are also some pitch variations that manifest themselves as subtle wow and flutter, along with a tendency for the pitch to decrease as the side progresses. This indicates that the cutter speed would have been overly sensitive to the mechanical loading of the cutter head on the disc, speeding up as the rotational load decreases as the head nears the centre of the disc. There is little that I can do to correct this without spending thousands of pounds on software that I would rarely use.
Nonetheless, it is a very good performance of this famous piece and I decided to bite the proverbial bullet and see what I could do with it. After putting the raw transfer through ClickRepair a couple of times, it was sounding pretty good, and with some careful distortion and rumble cancellation (by tweaking the stereo balance before mixing down to mono), I was able to get quite a passable result compared to what I started with.
You might also find that the noise floor also rises quite suddenly half-way through the third movement. I have no explanation for this. Perhaps it is due to some form of surface wear? Maybe someone dropped a worn osmium pickup onto this disc at some point, although there is no detectable increase in distortion which would be present if this was the case. I am sure that most of the noise is from the original master tape, because the noise floor drops quite significantly between each movement where the tape would have been edited. For this reason, I have not elected to use any wide-band noise reduction on this disc, as it would be a nightmare to implement with such a non-uniform noise profile. As a general rule, I do not apply it unless the SNR is lower than about 50dB or so. I did notice that the noise floor had a somewhat triangular nature, in that it increased quite rapidly above 5kHz, peaking at 12kHz and then rolling off above that point with what appeared to be a second order slope. This did not sound nice, and as I could not find a whiff of any musically relevant material above 7kHz, I chose to implement a first order roll-off above 8kHz to tame this effect to quite a satisfying degree.
7 inch EPs were notorious for being cut at a very low level in comparison to their 3-minute-pop-standard siblings, and this example is no different. Rumble was surprisingly high in relation to the recording level, and I had to be very meticulous in my search for the lowest note, a 46Hz sympathetic resonance that appears towards the end of the first movement. This is important to know, as it allows one to adjust the cut-off point of the rumble filter so as not to throw away anything musically relevant, while still attenuating the low frequency noise as much as possible. After I had applied this, the rumble was reduced to a fairly acceptable level, so I committed myself not to use DeNoise LF on this transfer. Interestingly, the low sympathetic resonance at F#1 (46Hz) would not have been present had the piece been played on the fortepiano of Beethoven's day, as its lower range was limited to two octaves below middle C (65Hz).
This recording is over 50 years old at the time of upload, and is therefore out of mechanical copyright. It is therefore available for free download.
Vorbis is used for site downloads as it provides transparency at about a third of the file size compared to MP3. There isn't really much point for using FLAC as the final listening format (all processing is done losslessly, of course) as the quality of the recordings themselves is already rather limited given their age. The general consensus on HydrogenAudio is that Q5 is enough for transparency with modern recordings, so the downloads offered are encoded at a more than ample Q7. Vorbis is by far a superior codec to MP3, as transparency is obtained at almost half the file size.
The audio tracks listed on this page are digital restorations of 78 RPM records in my possession, whose mechanical copyright has expired before the time of this pages publication. No later release is used so any copyright affecting such a release does not apply to any of the sound recordings shown on this page. Claims to the contrary may be vexatious if pursued. Any communication between parties claiming copyright of the material on this website and the author of this site will be published immediately with great derision. The contents of this page must not be copied represented or sold without express permission.
Michael Fearnley 2017