Restoration of 78 rpm records
The Process
When I was 13 my family moved back to Calne to a newly built house "Linden Field" in Horsebrook Park, after seven years in Devizes. By now I was trusted to use the family record player and allowed access to my father's collection of Jazz and Classical 78rpm records. About half of the Jazz records were Duke Ellington and catalogued as such in the exercise book where the collection was numbered and listed. The rest were catalogued in no particular order following on the Ellingtons. I got to know and love many of the records especially the Ellingtons but was not so taken with those that seemed a little too close to the dreaded "Trad" style.
Electronics
In the late 1960s I taught myself electronics and as a by-product of the job I then had, learnt how to build the sections of an amplifier from the inputs, cartridge,microphone etc, to power amps for loudspeakers. It occurred to me to try to process the 78s and see what I could do. Early efforts were interesting but unsatisfying and even with the help of a brilliant circuit designer, Bill Moore, progress was slow. Recently the acquisition of digital processing in the form of Audio Cleaning by both Magix and Oom has led to a standard that I am happy with and which, I hope, will find favour more widely.
Technical - Processing
The discs are played on a Thorens deck with a changeable cartridge at 78rpm as indicated by the strobe. Styli are selected to produce a suitable output.Styli have been provided by Expert Pickups of Ashtead, Surrey. The "front-end" has a high-frequency roll-off well above any frequencies likely to be found on these discs. The following stages include a bass lift which is adjusted to suit the material, 3 db points between 250 Hz and 500 Hz are typical. The material is then recorded digitally and processed digitally using Magix "Audio Cleaning" and X-OOM noise-reduction programmes.The processing does not remove all the background noise (comprising various degrees of crackle and hiss) but is intended to reduce it to be almost inaudible at a normal listening level. If you put your ear close to the tweeter you will hear the remnants of the noise and there may be tracks where I have been too cautious or too brutal.
Technical - Disc Faults
Some of the discs have had a carefree life and the results are not as I would wish. Some blemishes are almost impossible to overcome and they are still present to a degree. Other blemishes are in the recording and these are mentioned in the notes. The Brahms clarinet quintet slow movement has a thud from somewhere in the studio!. The Ellington recording "Reminiscing in Tempo" has some Wow which I thought might be my fault but the same defect is audible on my LP version. Some Wow is inevitable when the central hole gets widened with wear and the discs also tend to dish a bit.
The results
This website presents a portion of my father's collection together with a few Jazz 78s of my own and some classical pieces of my father's and also some from a large collection donated to me by a colleague. The Jazz sections, page by page, are about a CD's worth of music each and can be played track by track, illustrated by extracts from my father's exercise book where possible. The individual tracks are accompanied by recording data which I hope is correct! You can also play the whole "CD" by clicking on the All the tracks link.
Examples of the processing
The inside of a leather bound Schubert record sleeve
The oscilloscope I use
Some electronics
A Decca label record on my player
1. No noise reduction
Here is the opening of the Brahms Clarinet Quintet played with equalisation only , the same equalisation as in the subsequent examples
2. Some noise reduction
- The disc played again, this time with the electronic (real time) processing and the equalisation as before.
3. Final Result
The result of the digital processing following the electronic processing.
Comparison of transcriptions
The members of the GoodMusicGuide forum on the thread entitled "Great Recordings" were given the opportunity to compare various versions of the opening section of Artur Schnabel's recording of the Allegretto from Beethoven's "Moonlight" Sonata. The suggestion has been made that reducing surface noise, whether hiss or crackle, necessarily reduces the musical content of the recording particularly in the upper frequencies. I have collated the five selections and added my own to make a sixth. The samples were supplied and the processes identified by the distinguished member of the forum, george , as "Dante, EMI, Naxos, Nuovo Era and Pearl" to which CHM has been added. The loudness levels of the selections are roughly the same.You can hear the six repetitions of the music here.
Six transcriptions
I then subjected the whole compilation to a filter of 1/10 octave at 4 Khz, twice. This selects predominantly the tones of the top octave of the piano, i.e. overtones of the actual music, since the notes played are less than two octaves above middle C ( less than 1 Khz). You can hear the results below, substantially abbreviated. You can draw your own conclusions but it seems to me that the amount of music heard in the louder sections does not differ much. The third and fifth samples "Naxos" and "Pearl" have significant amounts of hiss compared to the other four samples. Whether there is more music to compensate is irrelevant if you prefer the overall quality of these particular transcriptions. The levels of the upper frequencies in the quieter moments ( the almost inaudible sections of the filtered sounds) is at least 40 db (>100 times) lower than the actual music suggesting that these overtones do not contribute much to our experience. Nevertheless the ear can be an extraordinarily accurate instrument and no one can hear with someone else's ears!
