Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Digital Restoration of 78rpm disks (Read 18970 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Digital Restoration of 78rpm disks

Is there any interest in digital restoration of historical recordings here? I have done some transfers that I think are pretty successful. I'd be happy to share them if anyone is interested.

Digital Restoration of 78rpm disks

Reply #1 –
Good thing you asked first, otherwise it'd be a TOS#9 violation.

Yes, of course there are people here who'd be interested, including me. But no, you can't post about them here, because there's a risk that you'd be exposing HydrogenAudio's admins and ISPs to legal troubles.

The only way would be if the audio on the records were in the public domain. But almost without exception, published recordings (being on a 78 almost certainly means it's published) won't be public domain in the US until 2067—and that's only if Congress's corporate masters don't extend the copyright term again.

Further reading: here (click). Page 20 is where they start talking about the state laws that cover pre-1972 recordings. And the Internet being global, all some copyright owner has to do is say you were distributing recordings to people in whatever state they want to sue or bring criminal charges in, and that HydrogenAudio was facilitating/supporting this, and any/all of the parties could be on the hook for it. Any arguments of it being "archival", "educational", "for personal use only", "fair use", etc., wouldn't matter... HydrogenAudio can't afford the risk, so it does not encourage or condone any posting, linking to, or mentioning places to get potentially copyright-infringing audio.

That said, very short clips for the sole purpose of demonstrating problems with audio codecs are tolerated, even though I believe that if some lawyer wanted to be a dick about it, they certainly could.

Digital Restoration of 78rpm disks

Reply #2 –
Quote from: mjb2006 on
HydrogenAudio can't afford the risk, so it does not encourage or condone any posting, linking to, or mentioning places to get potentially copyright-infringing audio.
That's most of YouTube out then!

Yes, several of us here are interested in audio restoration, and I sometime restore 78s. If you post samples, expect people to be interested in how you created them.

There have been other 78 threads, e.g.
http://www.hydrogenaudio.org/forums/index....showtopic=93659
http://www.hydrogenaudio.org/forums/index....showtopic=71317

There are also many threads about copying/restoring vinyl/LPs.

Cheers,
David.

P.S. I would only post 30 second samples directly here (unless out of all copyright everywhere) because those are the rules. However, I wouldn't think twice about posting full versions elsewhere - recording copyright is 50 years in the UK, and even in the USA historic material without great commercial value never seems to worry people. People usually post a caveat like this...
http://forum.talkingmachine.info/viewtopic.php?f=3&t=253

Digital Restoration of 78rpm disks

Reply #3 –
The copyright applies to the composition, which in classical music and early jazz all falls before 1924. The performance rights all fall before the 1974 copyright act. The law can vary from state to state in the US, but this stuff has all been released on PD releases. If folks are going to go all netcop on it, I'm not going to waste my time arguing over copyright law. I'd be happy to post MP3s and discuss how I accomplished the restoration. I'm also interested in discussing the works and performers too.

I'll start by posting a link to my restoration of Beethoven's Diabelli Variations by Artur Schnabel recorded in 1938. Schnabel was one of the greatest pianists who ever lived, but his recordings are notoriously difficult to restore. If you've heard the Pearl, EMI, Naxos Historical or the dozens of other releases of these recordings, you know that sound quality is all over the map, varying from bacon and eggs crackle to muffled like under a rug. I chose these recordings to experiment on because of their difficulty, and ended spending 5 or 6 months restoring, throwing the whole thing out and starting over three times. Finally, I arrived at a technique that peeled off the layer of noise without affecting the music underneath.

Anyone interested?

Digital Restoration of 78rpm disks

Reply #4 –
Quote from: bigshot on
I'm not going to waste my time arguing over copyright law.

Yet you're willing to make unsupported statements about the recordings being public domain, and you're willing to put this forum and its administrators at risk.

I chose my words carefully, talking about the audio, not the underlying composition. Read that report I linked to. Almost without exception, the audio in pre-1972 recordings is protected by state and common law.

Digital Restoration of 78rpm disks

Reply #5 –
If you aren't able to demonstrate that your offerings are completely free of copyright then you are not allowed to use this forum to share clips that are longer than 30 seconds.

Now that the rule has been questioned, discussed and presented, it is expected to be followed.  If you can't follow it then don't participate here.  If you disregard this post and continue to do so anyway you will be banned.  Yes, I will be nothing less than a complete dick abhout this.

Digital Restoration of 78rpm disks

Reply #6 –
It's all PD,  but feel free to not make links available.

My Facebook profile is under my name, Stephen Worth. Check out my photo albums for samples. These albums are public, so anyone can access them. Feel free to friend me if you want. If you want to hear how I accomplished my restorations, feel free to ask here. Let me know what you are interested in and I'll give you more discographical and technical info.

Digital Restoration of 78rpm disks

Reply #7 –
Quote from: mjb2006 on
Quote from: bigshot on
I'm not going to waste my time arguing over copyright law.
Yet you're willing to make unsupported statements about the recordings being public domain
It was not unsupported. He stated the composer is Beethoven (sufficiently long dead the world over to be out of copyright), and the recording date is 1938 (out of copyright 24 years ago across the whole of several continents, covering a few billion people).

Samples should be restricted to 30 seconds here, simply because of the confusing state of USA copyright law and strictly sticking to HA's TOS.

However, across the whole of Europe, Africa and Asia these recordings are out of copyright and anyone can do whatever they wish with them. People legally re-sell them, as also mentioned by bigshot. So it's a little unfair to say he made "unsupported statements".

Cheers,
David.

P.S. limiting samples to 30 seconds is merely a convention that, it is hoped, can be justified under the US defence of "fair-use". It does not respect international copyright laws - many other countries have no such fair-use defence in law. But it is in the TOS that you signed up for here.

P.P.S. Steven Stephen - you are lucky to be only one of two Steven Worths on facebook. I am virtually unfindable on there - which thinking about it, is probably even more lucky. For my tastes, your declicking leaves a little too many clicks/crackles in there - but then, I didn't hear the originals or what would have happened with higher settings. The results seem remarkably free from processing artefacts, and retain a nice, mostly undistorted sound. It's very listenable, and once you get used to it has very few distractions. How did you do it? EDIT: I hear some of your other examples are really excellent, so I guess that Beethoven recording was a real challenge.

btw, another relevant link...
http://www.hydrogenaudio.org/forums/index....howtopic=100114
...not something you can use today, but it's quite a hope for those recordings that fox current restoration techniques.

Digital Restoration of 78rpm disks

Reply #8 –
By the way, for those who'd like to hear the recordings we're discussing it's Stephen Worth (with a ph, not a v) at Facebook. The music is in my photo album.

Schnabel's Beethoven is the Mount Everest of restoration. The level of the music is quite low, a little bit muffled as if the lid of the piano was down, and because of wartime conditions in Europe, the shellac used was an ocean of crackle. Most restorations of these recordings involve big doses of broadband noise reduction, which removes the crackle, but takes out everything above the upper mids and blunts the dynamics by removing the attacks. This is what creates that xylophone like "Dresden China" sound that plagues a lot of historical recordings of piano.

I discovered by looking at the waveform that the noise in the Schnabel disks that it consisted of masses of individual clicks, not overall noise. So I cranked up a digital declicker to eliminate most of the clicks. When that was done, I did a tiny bit of pattern based broadband noise reduction to quieten the surface noise a bit, and to round off the jagged edges left by the declicker. The nice thing about impulse noise reduction is that it only applies noise reduction to the tiny fraction of a second when the click occurs. All of the sound around it is untouched.

Schnabel's recording of the Diabellis is one of the greatest piano recordings ever made. The PD versions that are available on CD or digital download, like the 99 cent MP3 box set at Amazon, mostly sound like you're listening to it wearing earmuffs. The EMI release is harsh and brittle with lots and lots of crackles. No one (except for Mark Obert Thorne's restoration for Naxos' PD Historical line) comes close to being as listenable as mine. I think a lot of restoration of 78s is done with a production line "one size fits all" approach. I find that every record is a little different.

Here is a PD release of this at Amazon...

http://www.amazon.com/Diabelli-Variations-.../dp/B004CE8CI2/

Now, you can't tell much from a 30 second clip, but notice how compressed the sound is. You can't feel the attack on the notes at all and the notes sound all evened out, like a xylophone. All of the touch is gone. The pitch of the transfer is WAY off too, with everything sounding like chipmunks.

By the way, 78s should always be listened to on speakers. Speakers present the natural balance of frequencies, and headphones have a tendency to over emphasize noise. I've found that headphones are useful for the first couple of passes of restoration to identify problems, but when you start to get close to being finished, you need to be auditioning on speakers... especially with frequency response, which is in many ways even more important to getting the sound right than noise reduction.

I have some examples of restoration of records that are about a century old. May I post those here, or will the copyright police get all frothy over that too?

Digital Restoration of 78rpm disks

Reply #9 –
My context:
For records where the original "masters" are lost, and where pristine disks are not to be found. Assuming that the content is important enough that someone is willing to spend some time, money and resources on generating improved-quality versions.

Did anyone try to align multiple instances of the same record? If two or more records can be assumed to be essentially identical except from short clicks, then a smart combination of them ought to have lots of potential?

-k

Digital Restoration of 78rpm disks

Reply #10 –
Quote from: knutinh on
For records where the original "masters" are lost, and where pristine disks are not to be found.


In this case, the master elements were the metal parts. Unfortunately, although the metal parts might theoretically contain more sound information than pressings, the sheer number of copies produced from the metal parts on a classic recording like this pretty much guarantees that the metal parts are worn when they are used for transfers today. Presumably, the EMI CD release of this is from the metal parts, but it is shrill and has distortion in the loud passages. The pressing I made my transfer from was in excellent shape, and was an early pressing when the metal parts would be immaculate. The main problem was the quality of the shellac that the record was pressed on. Declickers and manual declicking worked wonders.

Quote from: knutinh on
Did anyone try to align multiple instances of the same record? If two or more records can be assumed to be essentially identical except from short clicks, then a smart combination of them ought to have lots of potential?


Whenever I do a transfer, I always transfer in stereo, so I have two separate channels to work with. The music is common to both channels, but the noise is either particular to one side or the other, or completely out of phase to the music. Digital restoration filters compare the two channels to determine what isn't common to both and isolate the noise that way. Also, once the restoration is complete, summing the two channels removes all of the out of phase material, making the end result sound much cleaner.

I find that a lot of CD releases of vintage recordings neglect this last step. They leave the transfer in stereo. Ripping CDs of vintage mono recordings to mono rather than stereo when I make the listening copies for my music server and iPod not only saves hard drive space, it often makes the recordings sound better too.

Digital Restoration of 78rpm disks

Reply #11 –
Quote from: 2Bdecided on
the recording date is 1938 (out of copyright 24 years ago across the whole of several continents, covering a few billion people).

I don't doubt that it's copyright-free somewhere in the world. But can't that be said of any work? TOS#9 exists because HA can get in trouble in those places where it's not, so don't put anyone at risk over it.

Digital Restoration of 78rpm disks

Reply #12 –
Quote from: bigshot on
78s should always be listened to on speakers. Speakers present the natural balance of frequencies, and headphones have a tendency to over emphasize noise.

...not touching that one

Quote from: bigshot on
I find that a lot of CD releases of vintage recordings neglect this last step. They leave the transfer in stereo.

I, too, prefer the sound of true single-channel mono, but it's jarringly "distant" when shuffle-play places such a recording immediately after something in stereo, even if the side channel is just noise and drift. For this reason, and since I no longer am low on disk space, I stopped putting mono material ripped from CD into single-channel files, and now just let it stay in its more "present" form, wrong as it may be.

There have been other threads on here about dealing with mono source material on stereo recordings. IIRC, it has been mentioned that sometimes one channel is noisier than the other, so it's not always ideal to mix the two together, as opposed to just discarding the one you don't want. I'd be interested to know if there's a science to determining when it's better do to that, or if there's a way to combine the two techniques.

Digital Restoration of 78rpm disks

Reply #13 –
On speakers, the difference between stereo and mono isn't particularly jarring. Mono in mono always sounds better because you get a clear soundstage, not one that flickers in and out with phase errors. Also, if you have a 5:1 setup and use a DSP that processes the mains to create a surround environment, it works much better with true mono. If it is split with slight differences of noise, the DSP will pull the noise out and accent it in the rear channels. I really think that the future of historical recordings lies in creating multichannel abmbiences. There were some experiments with that in the early days of multichannel, but they used a weird format that no one had, and the stereo downmix was a muddled mess.

I've found that it's always best to work with disks that are in pretty good shape. When one side of the groove wall is beat up more than the other, it's usually caused by a mistracking acoustic soundbox. If that is the case, it's inevitable that either the inner grooves or the outer grooves are beat up on both sides. Doing a split mono, except in the cases of momentary groove damage, is usually salvage time.

Digital Restoration of 78rpm disks

Reply #14 –
Quote from: bigshot on
By the way, 78s should always be listened to on speakers.
Isn't that a bit annoying for everyone else on the train?!

Quote from: knutinh on
Did anyone try to align multiple instances of the same record? If two or more records can be assumed to be essentially identical except from short clicks, then a smart combination of them ought to have lots of potential?
I posted a link to a thread about that further up.

Quote from: mjb2006 on
Quote from: 2Bdecided on
the recording date is 1938 (out of copyright 24 years ago across the whole of several continents, covering a few billion people).

I don't doubt that it's copyright-free somewhere in the world. But can't that be said of any work? TOS#9 exists because HA can get in trouble in those places where it's not, so don't put anyone at risk over it.
We don't want to put HA at risk for anything. I'm just pointing out that most of the world has a different law on this from the USA.

Cheers,
David.

Digital Restoration of 78rpm disks

Reply #15 –
Quote from: 2Bdecided on
Quote from: bigshot on
By the way, 78s should always be listened to on speakers.
Isn't that a bit annoying for everyone else on the train?!


You have to remember that everyone in the 78 era listened to music on speakers. Few people had headphones until the LP era.

Acoustic gramophones incorporate psycho-acoustic principles to make music sound good that are completely different than modern sound reproduction. They used horns and advised to place the phonograph in a corner where the walls extended the effectiveness of the horns. They also played around with totally unbalanced, but totally musical frequency responses. None of this comes across well on headphones. It's really difficult to recreate on modern electrical transcriptions too. I've done some experimentation with trying to transfer 100 year old recordings to make them sound the way they do on an acoustic gramophone. When I get a chance, I'll post some of my experiments here. (There isn't a place on earth where pre-1924 recordings aren't public domain.)

Digital Restoration of 78rpm disks

Reply #16 –
Quote from: bigshot on
(There isn't a place on earth where pre-1924 recordings aren't public domain.)

(You keep insisting on this, but you are wrong. In the U.S., everything published prior to 1923 is in the public domain, except sound recordings. All sound recordings published prior to Feb. 1972, including pre-1923 recordings, are under state protection until 2067. This is a known issue and is a significant topic covered in the document I linked to. As explained in that paper, there is discussion of bringing pre-1972 recordings under the scope of federal copyright law, which would result in the pre-1923 recordings immediately entering the public domain, although the Copyright Office recommends a 3-year delay.)

(Hopefully using parentheses makes this whole conversation not count.)

Digital Restoration of 78rpm disks

Reply #17 –
Quote
On speakers, the difference between stereo and mono isn't particularly jarring.

I do agree; when I said it was jarring, I had headphones in mind, of course. Whoops.

Quote from:  link=msg=834172 date=0
Acoustic gramophones incorporate psycho-acoustic principles to make music sound good that are completely different than modern sound reproduction. They used horns and advised to place the phonograph in a corner where the walls extended the effectiveness of the horns. They also played around with totally unbalanced, but totally musical frequency responses.

I'd be interested to read more about this, if you have any literature (online or off) to point to.

Digital Restoration of 78rpm disks

Reply #18 –
Quote from: mjb2006 on
Quote from:  link=msg=834172 date=0
Acoustic gramophones incorporate psycho-acoustic principles to make music sound good that are completely different than modern sound reproduction. They used horns and advised to place the phonograph in a corner where the walls extended the effectiveness of the horns. They also played around with totally unbalanced, but totally musical frequency responses.

I'd be interested to read more about this, if you have any literature (online or off) to point to.


Each manufacturer had its own acoustics lab, and had the ability to design the recordings to suit the phonographs and vice versa. Competition in the early days was cutthroat, so there isn't a lot of info about what went on in the labs. You have to piece it together by the evolution of the machines and records each company made. For instance, experimenting with different horn shapes at Victor led by chance to the development of a primitive exponential horn. Bell Labs later worked out the details and Victor incorporated it into their Orthophonic line.

Edison Laboratories documented their process the best. Here are a couple of great Edison photos. This first one is a recording session...



This one shows the acoustics lab. They appear to be experimenting with horn sizes and shapes here...



Some of the best sites on antique phonographs are...

MAPS http://antiquephonograph.org/
NIPPERHEAD http://www.nipperhead.com/
VICTOR-VICTROLA http://www.victor-victrola.com/index.html

If you would like to hear some of my restorations of 100 year old recordings, PM me. I'm happy to share them.

Digital Restoration of 78rpm disks

Reply #19 –
Quote from: bigshot on
Acoustic gramophones incorporate psycho-acoustic principles to make music sound good that are completely different than modern sound reproduction. They used horns and advised to place the phonograph in a corner where the walls extended the effectiveness of the horns. They also played around with totally unbalanced, but totally musical frequency responses. None of this comes across well on headphones. It's really difficult to recreate on modern electrical transcriptions too. I've done some experimentation with trying to transfer 100 year old recordings to make them sound the way they do on an acoustic gramophone. When I get a chance, I'll post some of my experiments here. (There isn't a place on earth where pre-1924 recordings aren't public domain.)
For acoustic recording practices, and the frequency responses of various electrical recording systems (not to mention 1000 other useful bits of information) it's well worth reading Peter Copeland's book...

http://forum.talkingmachine.info/viewtopic...f=8&t=12278


I don't buy the argument about totally unbalanced but totally musical frequency responses - sorry! Once you correctly compensate for the recording characteristic, electrical playback far surpasses anything possible with a gramophone in terms of transparency, realism, and revealing hidden details of the recording which are inaudible on period machines.

I once simulated the (measured) frequency response of my little HMV 101 and applied it to an electrical transfer - I was astonished at how much like (a recording of) the gramophone it sounded. However, the record itself sounded far better without this "gramophone" frequency response applied.

I guess far better results are possible with acoustic discs, and simulations of better acoustic gramophones. Or you could just do what Nimbus did, and mic up the best gramophone you can find...
http://www.youtube.com/watch?v=_MA957TfmIs
http://www.youtube.com/user/grahamrankin

Cheers,
David.

Digital Restoration of 78rpm disks

Reply #20 –
Quote from: 2Bdecided on
I don't buy the argument about totally unbalanced but totally musical frequency responses - sorry! Once you correctly compensate for the recording characteristic, electrical playback far surpasses anything possible with a gramophone in terms of transparency, realism, and revealing hidden details of the recording which are inaudible on period machines


First of all, I'm only talking about acoustic recordings here, not electrical ones. Electrical sound better with electrical transcription.

But for acoustic recordings, the only frequencies that benefit from electrical transcription are low frequencies, which the Mica soundbox is incapable of reproducing. But for the core frequencies, the sound that comes out of a grammophone is amazingly lifelike and present, and it is very difficult to capture that in an electrical transfer. If you transfer flat, it sounds noisy. If you transfer at perceived flat, the dynamics drain out. I've got acoustic phonographs (both mica soundbox and orthophonic) and I've done a lot of experiments with transferring this stuff, working to match the vivid sound of acoustic playback. It is a lot easier to get rid of noise than it is to get that clear, bright sound image. The soundbox swallowed noise and wolf tones and emphasized frequencies around the human voice. This, combined with the very close proximity of the performer to the recording horn makes for wonderful sound.

Whenever I play a Caruso record on my cabinet Victrola for friends, they are amazed. It projects an aural image into the room in front of the phonograph that is so real, it makes the hairs on the back of your neck stand up. I've never heard a CD transfer that comes anywhere close. The sound of an acoustic phonograph playing acoustic recordings is very unique, and is nothing like the way we record or playback music today.

Of course, I'd be happy to provide MP3 samples, but...

Digital Restoration of 78rpm disks

Reply #21 –
I posted "Liberty Bell March" by the Columbia Band from 1901 on my Facebook page yesterday. This is one of the earliest records in my collection.

Facebook > Stephen Worth > Photos > Albums > 1920s and 30s music > Liberty Bell March

Digital Restoration of 78rpm disks

Reply #22 –
Quote from: bigshot on
It's all PD,  but feel free to not make links available.

My Facebook profile is under my name, Stephen Worth. Check out my photo albums for samples. These albums are public, so anyone can access them. Feel free to friend me if you want. If you want to hear how I accomplished my restorations, feel free to ask here. Let me know what you are interested in and I'll give you more discographical and technical info.


Look for the profile photo of Tom Mix.

Digital Restoration of 78rpm disks

Reply #23 –
That's me, pardner!

Digital Restoration of 78rpm disks

Reply #24 –
Quote from: bigshot on
I posted "Liberty Bell March" by the Columbia Band from 1901 on my Facebook page yesterday. This is one of the earliest records in my collection.

Facebook > Stephen Worth > Photos > Albums > 1920s and 30s music > Liberty Bell March
Have you hidden your albums, or am I now looking at the wrong Stephen Worth on facebook? I can see one from Cincinnati, and another one who Studied at University of West Georgia, neither with open albums.

Cheers,
David.