Re-mastered 78's

[Andrew2]

Originally Posted by michamoo

It would be nice if someone would release a cleaned up recording of "Go now" by the Moody Blues as the last minute is and has always been very distorted. This has been discussed at length in many places on internet forums and it seems even the actual master recording is the same and no decent version of that recording ever existed!

Ah, that's interesting. This record was the first I'd gone out and bought with my own pocket money ( I was twelve, I think) and I assumed the distortion was just down to poor pressing. When CDs came along I was surprised to hear the distortion still present and thought it must have been an engineer being a bit liberal with the mic faders. Would we want a clean, distortion-free version? The distortion gives the song a really loud, urgent feel.

[Synchrodyne]

Given that the original question was about what was done (or not done) in the pre-digital era, some timeline information might be appropriate.

The consumer level noise blankers appear to date from 1977.

The SAE 5000 was announced, reviewed and advertised in Audio magazine 1977 June.

The Garrard MRM101 was mentioned as being forthcoming in Wireless World 1977 November p.57.

The first Packburn professional unit, the Model 101, was introduced in 1982, see: http://www.packburnelectronics.com/history/.


Cheers,

[wd40addict]

A modern real-time heir to the Garrard MRM exists, but is not cheap:

http://www.sweetvinyl.com

[martin.m]

Thanks to everyone who has replied. It seems the answer to my question is that record companies either had or looked for excellent copies of the 78rpm discs, then applied minimal analogue filtering.

Ted, you are clearly an expert in this field. I think I have one of those Columbia "silent surface" pressings. A very young Doris Day singing "Canadian Capers" with one of the 1940s Big Bands. My copy looks like new and when played on a 1950s radiogram (upgraded by fitting a Garrard SP25 turntable and ceramic cartridge) sounds nearly as good as vinyl.
Some record companies pressed vinyl 78s around 1958. I have one on the Pye label, Lonnie Donnegan's "Does your chewing gum lose it's flavour". My record has seen better days but still plays well without the hissing surface noise associated with most shellac discs. I was never impressed with the 45rpm singles from the 50s and 60s. Many sounded rough and distorted, especially as the record got near to the end. They improved in later years though. Chris Isaak's "Wicked Game" from 1990 sounds as good as a CD when played on a half decent turntable and magnetic cartridge.

[astro63]

i remember seeing a program back in the late 80s something about "nipper" the dog and they were transfering rare 78's onto reel 2 reel tape turning fast then during the playback everytime a click appeared they would splice the tape and cut the click out, now its computers and so much quicker

[G6Tanuki]

"De-clicking" of records is in some ways similar to the issue of reducing ignition-interference that plagued TV/radio-reception.

The fundamental issue is that a 'click' can drive electronic circuitry into saturation - coupled with the extremely fast rise-time you then have what amounts to a square-wave, and as the good Mr. Fourier showed us, that means a slew of harmonics - all of which can trigger 'ringing' in the signal path so the effects of the click can be noticed even after the click has ceased.

The simplest 'noise limiters' - things like back-to-back diodes - just clipped the peaks. Good ones used 'soft' clipping [the old 'westector' diodes were strangely good for this] to reduce the harmonic-generation.

Smarter 'blankers' rather than clipping the peak actually clamped the signal to zero during the click - essentially punching a 'hole of silence' into the audio for the duration of the click. The probloem was that, though this was less-noticeable than the click itself, there was still a level-discontinuity across the 'hole'.

There were various analog-attempts at using things like PAL-TV glass 64uS delay-lines so that when a 'hole' was created after a click had been removed, the audio from either side of the hole could be 'stretched' to fill in the gap. All with variable levels of success.

These days, a cheap PC, soundcard and suitable software can achieve things earlier restorers could only dream at.

My greatest 'annoyance' in restored/remastered audio is crude hiss-reduction, where in the silent/quiet periods the hiss is removed, but in passages where the volume's higher the hiss re-emerges. It's like imagining the band performing on a beach, with the waves coming in behind them when they play loud!

[martin.m]

The problem I have with most of my 78rpm records is not clicking but the continuous hissing surface noise of the shellac. I have Magix Audio Cleaning Lab 2005 for transferring vinyl to WAV files (they want the user to pay extra for converting to MP3). There is a de-clicker but I don't find it very effective. I have never tried removing background hash from 78s though. Perhaps other members have experience of this program.

[dseymo1]

Originally Posted by G6Tanuki

The fundamental issue is that a 'click' can drive electronic circuitry into saturation - coupled with the extremely fast rise-time you then have what amounts to a square-wave, and as the good Mr. Fourier showed us, that means a slew of harmonics - all of which can trigger 'ringing' in the signal path so the effects of the click can be noticed even after the click has ceased.

Quite true, but ringing can also be initiated before the signal even reaches the electronics: mechanical resonances of the arm / cartridge combination are excited by the clicks too.
It surprised me when I was first fortunate enough to own a high-quality turntable, arm and cartridge how noise-free reproduction was compared with run-of-the-mill systems.

[Ted Kendall]

Originally Posted by martin.m

The problem I have with most of my 78rpm records is not clicking but the continuous hissing surface noise of the shellac. I have Magix Audio Cleaning Lab 2005 for transferring vinyl to WAV files (they want the user to pay extra for converting to MP3). There is a de-clicker but I don't find it very effective. I have never tried removing background hash from 78s though. Perhaps other members have experience of this program.

I'm sorry to be damping about this, but nothing except CEDAR really works on shellac 78s, in my experience. Shellac crackle has far more events per second than even the worst vinyl, placing a premium on the accuracy of the modelling algorithm - as I've said. a simple blanker or any of the other devices of yore designed for vinyl clicks decend into a burbling mess when given crackle to deal with.

Hiss is difficult to treat, not least because it has a spectrum which is not white. The noise peaks in the low kHz region, and in some ways this is fortunate, because sufficiently good software (such as CEDAR NR-5) allows the operator effectively to equalise the surface noise, by reducing it in the frequency bands which are most troublesome whilst leaving the rest of the spectrum intact. Basic, broad band hiss reduction falls over with shellac noise - if you use enough to tackle the mid-range noise, all the low level information is sucked out of the LF region, where the reverberation and ambience live, giving rise to the disembodied "there's no there there" effect. It is as well to remember that all hiss reduction algorithms are essentially dynamic filters - careful design and use can render their applcation transparent, largely by exploiting the masking characteristics of human hearing, and only treating the most troublesome parts of the spectrum is a very effective masking technique, as Ray Dolby showed in his Dolby A system, which is completely inaudible in operation when properly aligned.

There is a lot to be said, even now, for picking up a Packburn, if you can find one - it really was quite a good instrument in its day for improving the music to muck ratio of shellac. It still arguably makes a better fist of it than cheap software, whose designers do not understand that shellac is different from vinyl in kind, not degree.

[bluepilot]

I'm sorry to be damping about this, but nothing except CEDAR really works on shellac 78s, in my experience.

CEDAR is undoubtedly an excellent piece of software and if you are a professional it probably makes sense to buy it. However for home use the price is completely OTT in my opinion. Personally I use Clickrepair: Clickrepair
which produces very good results for a reasonable price (no connection, just a satisfied customer). It does most of what I need to do. Every now and again I find a record with a stubborn click and it's much cheaper to buy a second copy of the record which doesn't have a click there than spend thousands on CEDAR. (One one occasion I discovered that an annoying thump was actually recorded on the record where apparently something had fallen on the floor during the recording).

Even CEDAR isn't infallible. For example I have a CD of Charlie Kunz piano medleys which boasts that it was processed with CEDAR and some of them are very noisy whilst others are comparatively quiet. Presumably it's a function of the original record which brings us back to an earlier comment that the most important thing is to get a good quality copy in the first place.

Having said that, retouch sounds like a useful function to have.

[Ted Kendall]

Originally Posted by bluepilot

CEDAR is undoubtedly an excellent piece of software and if you are a professional it probably makes sense to buy it. However for home use the price is completely OTT in my opinion.

Well of course it is. The fact remains that it is the tool of choice. My argument is that cheap DSP is expensive at the price, because unless the processed audio sounds as if it wasn't damaged in the first place, you're better off with a Packburn or something similar, even a second-hand obsolete CEDAR box, than the sound of duff software falling over. At least analogue processors fail gracefully rather than sounding like a robot with indigestion.

Originally Posted by bluepilot

Even CEDAR isn't infallible.

Well of course it isn't. It's a tool, not a panacea. Like any tool of sufficient power, it can completely louse up the sound if not properly used. I have the experience to drive my costly tools to their limits. Some don't. One dozy character did a whole album with the decrackle function set to "detect", which is only meant as an aid to adjustment. No surface noise - and precious little music, either.

[Grubhead]

Originally Posted by Synchrodyne

Given that the original question was about what was done (or not done) in the pre-digital era, some timeline information might be appropriate.

The consumer level noise blankers appear to date from 1977.

The SAE 5000 was announced, reviewed and advertised in Audio magazine 1977 June.

The Garrard MRM101 was mentioned as being forthcoming in Wireless

I remember seeing the Garrard MRM101 being demonstrated at the Hallam Towers Hotel Hi-Fi Show. I was very impressed with it. They had a very badly scratched record playing and you couldn't hear one click, till the chap turned down the control then it was shocking! He did point out that the record was an extreme case! There was a red light on the device which lit on detection of clicks. In the case of that record it was never off!
I was surprised that it never really caught on.

Originally Posted by martin.m

Thanks to everyone who has replied. It seems the answer to my question is that record companies either had or looked for excellent copies of the 78rpm discs, then applied minimal analogue filtering.

Ted, you are clearly an expert in this field. I think I have one of those Columbia "silent surface" pressings. A very young Doris Day singing "Canadian Capers" with one of the 1940s Big Bands. My copy looks like new and when played on a 1950s radiogram (upgraded by fitting a Garrard SP25 turntable and ceramic cartridge) sounds nearly as good as vinyl.

The Doris Day 78 dates to 1950 precisely for it entered the UK sales chart on the 15 January 1950 and reached number 2 and spent 23 weeks in the top 40. It sold over 200,000 in the UK.

[John10b]

Thank you Synchrodyne for the link to Packburn, I’ve just listened to some before and after recordings, wonderful stuff.
Cheers
John

[martin.m]

Graham Foster, an English Folk Dance enthusiast, has achieved excellent results cleaning up old 78rpm recordings of folk dance music using very modest equipment. There are some before and after samples on his website.

www.gfoster.info

[TIMTAPE]

Originally Posted by Ted Kendall

... It is as well to remember that all hiss reduction algorithms are essentially dynamic filters - careful design and use can render their applcation transparent, largely by exploiting the masking characteristics of human hearing, and only treating the most troublesome parts of the spectrum is a very effective masking technique, as Ray Dolby showed in his Dolby A system, which is completely inaudible in operation when properly aligned...

It's interesting you mention Dolby NR and its absence of artifacts when properly aligned. The same cant be said for any playback only Denoiser. Even the humble Dolby B could achieve 10 to 15db of genuine noise reduction in the treble area. That's 10 to 15db more than the fanciest CEDAR or Izotope Denoiser can do today. The old analog NR systems like Dolby, DBX, ADRES, ANRS, HiCom, Burwen could work well not because they were analog, or because they were old, but because they were double system, or companders. They pushed the programme above the system noise at the record end, the only place to do it. Then the playback decoding merely undid the compression.

Trying to expand down unwanted noise at the playback end is doomed to failure if unlike double system NR the recording was never intended to be expanded on playback. The result will never be transparent. All we can do is use it very sparingly and hope that not too many people notice the artifacts. It's not the fault of the tool. It just has an impossible task.

[Hartley118]

Originally Posted by Ted Kendall

Before the advent of such devices as the Packburn, getting a decent 78 transfer was all about the source, as opposed to today's position, where it is mostly about the source!.............................

The Laughing Policeman is an interesting case - it is a very early electrical recording, using the unmodified Western Electric system, which was very linear over its 5kHz bandwidth. The sound is remarkably natural and "present" - part of this is the 2.8 kHz peak in the WE microphone. The reason we can hear so far into the recording is that it was issued by Columbia, using their "silent surface" - a laminated pressing which really was as near as dammit noise-free - in good condition, it can nearly be taken for vinyl, although the different response of the pickup on a harder surface gives it away. I remember doing this for CD, and was bowled over by the sheer sense of Penrose standing there in the plane of the speakers. Marvellous record - with Ted Heath on trombone, too - what's not to like?.....

Those Columbia 'New Process' laminated pressings from the 1920s really were remarkable. As Ted says, the best Columbia early electric recordings using the Westrex system possess a 'presence' that in my view was rarely achieved in later 1930/40s 78s. Westrex discs can be identified by the impressed tiny letter 'W' in a circle and if you find one relatively unworn, it can be a revelation. Although their bandwidth is limited to 5kHz, the quiet surfaces allow a little judicious treble boost to brighten the sound. However, few of the 1920s recordings to be found nowadays are of the 'silent surface' quality, most being later conventional crackly shellac pressings

By 1930 Columbia and HMV had merged under the EMI umbrella and a new improved recording system was developed in EMI labs by one Alan Blumlein. Developed mainly to avoid paying licensing fees to Western Electric, this innovative system also had the advantage of a wider bandwidth - up to 8kHz I understand. Unfortunately, the merger also spelt the end of the New Process laminated quiet pressings. Quite why they were killed has always been a mystery to me because their audio quality was way ahead of its time. Perhaps they were too costly, or maybe they weren't suited to the mass-market HMV acoustic gramophones: the abrasive content in conventional shellac pressings was said to increase record life by sacrificing the needle rather than the record. Unfortunately it produced that familiar 'frying bacon' background that we took for granted until technology improved in the mid-1950s.

A Blumlein Columbia disc is identified with an impressed tiny letter 'C' in a circle. It would be great to hear one on a quiet laminated pressing, but I've never managed to find one. Maybe there aren't any.

Martin

['LIVEWIRE?']

No doubt the best quality sound produced on 78rpm discs in the mid to late 1950s was that emanating from the vinyl 78s issued by the Pye group (Pye-Nixa and Mercury primarily) These had a virtually silent surface comparable to the 45s & LPs of the day, and a good condition copy of one of those pressings will still to this day have a very good S/N ratio. Not sure what the frequency response of vinyl 78s was, though it probably extended to at least 10kHz. Anyone know the exact figures?

[Phil G4SPZ]

I listen to a lot of pre-war music sourced from digitally-enhanced re-mastered 78s, most of which has been very well engineered and I enjoy it immensely. However I also enjoy occasionally playing some of my own 78s on various reproducers, including mechanical ones. Somehow the associated surface noise is just part of the experience, and was the type of sound that people would have heard from records, certainly up to the 1950s.

[Synchrodyne]

Originally Posted by 'LIVEWIRE?'

Not sure what the frequency response of vinyl 78s was, though it probably extended to at least 10kHz. Anyone know the exact figures?

Well, Decca claimed 14 kHz for its ffrr recordings, presumably of the shellac 78 rev/min type, back in 1947:



So I'd expect no less from the vinyl 78 type.

Cheers,

[Ted Kendall]

Originally Posted by 'LIVEWIRE?'

No doubt the best quality sound produced on 78rpm discs in the mid to late 1950s was that emanating from the vinyl 78s issued by the Pye group (Pye-Nixa and Mercury primarily) These had a virtually silent surface comparable to the 45s & LPs of the day, and a good condition copy of one of those pressings will still to this day have a very good S/N ratio. Not sure what the frequency response of vinyl 78s was, though it probably extended to at least 10kHz. Anyone know the exact figures?

The last 78s were cut on kit indistinguishable from that used for LP, and even had a similar EQ curve - the British Standard was 450/50uS. Chappell and other prodcution music companies continued to make vinyl 78s with RIAA EQ for commercial use into the seventies - the high disc spedd and coarser groove pich made on-air cueing easier.