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: Cleaning up 78 transfers (Read 55642 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Cleaning up 78 transfers

Reply #25
Ultimately your ears are the best judge.

That's exactly what I'm finding.
Guide my thinking here.... In understanding the RIAA inverting, is it true that every device that I can use today to capture vinyl will have RIAA equalization built into it and therefore will have to have those values inverted? Or is it that the devices are the same as before but the old recordings were mastered with a certain EQ (specifically 78s)?

And finally, what is being lost by not inverting the RIAA curve? I don't seem to understand that. The 78s I capture sound better without it inverted. Unless my ear is wrong which is totally possible.

Cleaning up 78 transfers

Reply #26
Except the oldest, the 78's had some EQ applied and should have  some EQ applied upon playback. That EQ is not the RIAA curve, but the RIAA curve may very well (for all that I know) be better than nothing.

Cleaning up 78 transfers

Reply #27
Ultimately your ears are the best judge.

That's exactly what I'm finding.
Guide my thinking here.... In understanding the RIAA inverting, is it true that every device that I can use today to capture vinyl will have RIAA equalization built into it and therefore will have to have those values inverted? Or is it that the devices are the same as before but the old recordings were mastered with a certain EQ (specifically 78s)?

And finally, what is being lost by not inverting the RIAA curve? I don't seem to understand that. The 78s I capture sound better without it inverted. Unless my ear is wrong which is totally possible.

Yes, pretty much any electrical device intended for vinyl playback will have the RIAA eq built in. In crude terms the bass is reduced and the treble increased when the disc is cut and the bass is boosted and the treble reduced on playback. So, you should end up with the original tonal balance but you have reduced noise (which is more noticeable at higher frequencies) along with the "enhanced" treble. I believe The RIAA curve became the standard in the late 50's.

The problem for 78 replay is that they pre-date vinyl and there was no standard in place. Each company did what they thought best - this wiki article tells you more. So, if you play back a 78 on equipment intended for vinyl replay the wrong inverting eq curve will be applied. If it matters to you that much you may well be able to find out what eq was applied to your records during manufacture (try Google). You can then reverse the RIAA eq and apply the correct eq or by comparing the eq used in manufacture to the RIAA curve, have a stab at correcting the difference to try getting back to how it should be. Some people try to avoid all of that by recoding flat - usually via mic preamps - and then applying the correct eq. Personally, I find that all a bit too much trouble and I just use the RIAA eq'd capture as a base and adjust by ear to achieve what sounds good to me. It's probably a million miles away from "accurate" but 78's aren't exactly hi-fi to start with so I don't worry about it.

If you don't invert the RIAA eq your recordings will have a bass lift - often desirable, as long as it isn't overdone - and less treble, which in the case of 78's is where a lot of the noise is most apparrent, which is also desirable. That may be why you prefer them without the RIAA being reversed


Cleaning up 78 transfers

Reply #29
Just so you know, there are more than a few phono preamps for 78s on the market. Some will provide just about every, if not every, EQ ever used on commercial pressings, some are limited to just the most common EQs.

Cleaning up 78 transfers

Reply #30
Just so you know, there are more than a few phono preamps for 78s on the market. Some will provide just about every, if not every, EQ ever used on commercial pressings, some are limited to just the most common EQs.
Thanks! That's really cool but I couldn't BEGIN to discern which brands were good, how to compare one from the other, etc. Plus if the curves can be applied digitally in Sound Forge or Audacity or something, it would be cheaper for me.

Cleaning up 78 transfers

Reply #31
Check this out though: http://wiki.audacityteam.org/wiki/78rpm_playback_curves Theres a shitload of de-emphasis curves for 78rpm's.

I have no idea how to take the values of these curves and plot them on a graph to create my own curves to apply to a recording. Can anyone help?


I don't use this feature, but ... http://forum.audacityteam.org/viewtopic.php?f=39&t=35665 ? http://forum.audacityteam.org/viewtopic.php?f=28&t=33824 ?

Cleaning up 78 transfers

Reply #32
Wow... gets pretty intense. Ultimately, these are 78s. And once the RIAA curve is reversed (if its even deemed necessary) an understanding of the music, how it was meant to sound, etc and getting it to the point one wants might be the only important thing, I'm thinking. I don't know. I'd love to say: RIAA done, now it's a Decca so I apply the Decca curve, and that's that.

 

Cleaning up 78 transfers

Reply #33
Except the oldest, the 78's had some EQ applied and should have  some EQ applied upon playback. That EQ is not the RIAA curve, but the RIAA curve may very well (for all that I know) be better than nothing.
It's a lot closer than nothing! For most records in most frequency ranges at least. There are exceptions.

Cheers,
David.

Cleaning up 78 transfers

Reply #34
You didn't follow the links posted earlier...

From http://www.rfwilmut.clara.net/repro78/repro.html
Quote
Not many people have suitable preamplifiers: it is possible to transfer 78s using the usual RIAA equalization for LPs and then convert. Equalizer by Brian Davies is free and will convert RIAA or flat (unequalized) to a number of other equalizations. If you have transferred a 78 at a lower speed and then speeded it up in an audio editor it will also adjust the equalization to suit. Prior to this being available I worked out some figures, and I have left them posted here in case anyone wants to experiment with them.


You can't just remove the RIAA curve and leave it at that. You'll be worse than you started.

Cheers,
David.

Cleaning up 78 transfers

Reply #35
You can't just remove the RIAA curve and leave it at that. You'll be worse than you started.

Ok, and I'm reading about the curves but, for example, I'm doing some Capitol 78s from the 40's. What curve do I apply for those? I have a list of tons of curves but I don't understand how to plot those figures onto a chart to save that curve?

Cleaning up 78 transfers

Reply #36
Quote
I'm doing some Capitol 78s from the 40's. What curve do I apply for those?
I just don't think it's that critical (or optimal) to find the original EQ curve.    I think you'll get the best results by ear.

What's your actual goal?  Do you want the best sound (I'd be going for the best sound to my ear).  Or, do you want to accurately duplicate what someone would have heard on their equipment in the 40s?  i.e. What the record producer expected/wanted the listener to hear?

Think about the equipment it was designed to be played on.  A mono tube-amp, probably with a single "full range" speaker (no woofers or tweeters).    This was before "high fidelity", and I assume the record production goal was to get the best "vocal clarity" on equpment of the day.  If you perfectly reverse the recording curve, it's probably not going to sound that good, especially on modern full-range equipment. 

The older 78s were played on gramaphones with an acoustic horn used as a speaker/amplifier.  If you want to duplicate that sound, your EQ curve would have to simulate the (unknown) acoustics of that set-up.  But, that might not give you the best sound.    Inverting the recording-curve and playing through a modern more-flat system probably won't give the best results either.   

So, I'd start with RIAA and adjust by-ear from there.    (I know some people don't like to mess with the original sound, but I sometimes add a little high-end to older dull-sounding LPs, because it sounds better to me than the "correct" RIAA equalization.)

Cleaning up 78 transfers

Reply #37
Now remember that the OP did the recording at 45rpm, which means that the RIAA EQ curve will be moved some three quarters of an octave up when the speed is adjusted back up. (BTW, as in a reference I left, it isn't so that '78' was fixed at 78!)

Cleaning up 78 transfers

Reply #38
So, I'd start with RIAA and adjust by-ear from there.    (I know some people don't like to mess with the original sound, but I sometimes add a little high-end to older dull-sounding LPs, because it sounds better to me than the "correct" RIAA equalization.)

I've been reading for about 2 hours about all this and about curves (still not understanding how they're applied to an audio capture of a 78) and I think you're 100% correct. I think that...
a) trying to replicate what a completely analog phonograph horn sounds like is pointless. You can't replicate that sound created by a reproducer and vibrating through a large wooden case. I know, cause I have one and that sound is a unique thing that is the reason for owning a phonograph.
b) music of the day would have been mixed to sound good on an analog system and probably not what the industry typically balances for today (crappy iphone headphones and car stereos)
c) this kind of music restoration doesn't happen as often for this very reason. It doesn't JUST require a sound engineer but also a certain degree of knowledge of the time period and what music was meant to sound like.
To a certain degree, taking 78s that played on a phonograph and getting them to be listenable today through speakers is going to involve some improvisation if, like me, you want to reproduce an accurate sound and not necessarily a good remastered sound. Sorry don't know how else to put it. It's a bit to me like when a classic film is released with a new (and ridiculous) 5.1 mix when there was only ever center channel mono in the theaters of the time.
Now, how about the other point though, that sometimes I feel like the RIAA inverted makes the recording sound worse? Wouldn't it be best then to leave it alone without inverting the RIAA? On some really beaten up 78s it just accentuates the surface noise and I feel like it's killing any lower end that's there to a point that makes the whole song lose fidelity.

I appreciate the thoughts. I'm transferring a lot of my own 78s. Some of them really valuable cause I want to be able to listen to them in a row, as albums and saving the actual record for times when I really wanna hear my 1927 phonograph in my apartment.

Cleaning up 78 transfers

Reply #39
You should of course keep the original file (both before and after 'cleaning up' ... the 'before' in case you screw up something during the process), so a good start would be to keep the rip with the RIAA curve (and if you do any speed adjustment but 78/45ths, then tag properly or keep a backup!), and you could always make adjustments later if the sound starts to annoy you.

A hint if you store in lossless formats: you could use one format for 'unprocessed' and one for 'processed'. Tags etc. may be accidentally deleted, as long as you stick to e.g. WavPack for unprocessed and FLAC for processed, you will always be able to tell which one was which.

Cleaning up 78 transfers

Reply #40
You should of course keep the original file (both before and after 'cleaning up' ... the 'before' in case you screw up something during the process)

I always keep a pretty strict chain of files while I work always keeping the one 'clean' capture and then adding my own lettering system as I go so I can jump to any level whenever I want. I use WAVs all the way. When I'm done I 'master' a WAV as well. Then, if I decide I'm happy, I keep the first capture, the cleaned up master, and a FLAC for listening.

Cleaning up 78 transfers

Reply #41
You can't just remove the RIAA curve and leave it at that. You'll be worse than you started.

Ok, and I'm reading about the curves but, for example, I'm doing some Capitol 78s from the 40's. What curve do I apply for those? I have a list of tons of curves but I don't understand how to plot those figures onto a chart to save that curve?

I'm going to post this link for the third time. If you don't follow it this time, I'm giving up!

http://www.clickrepair.net/software_download/equalizer.html

The only gotchas I found were that you may have to reduce the gain yourself manually (there's a setting, but it won't set it automatically to avoid clipping, so you'll have to set it yourself), and (on my system at least) you seem to have to pick the curves you want before loading the file - picking them after loading the file sometimes causes the previous set of curves to be applied instead. Which is very confusing!

Which ones to use? The choice between Blumlein and Westrex (at least for UK 78s) is explained here...
http://www.rfwilmut.clara.net/repro78/repro.html#eq
Quote
78s: WESTREX (English Western Electric) for HMV 78s with triangle matrix code and English Columbias with a (w) matrix code. BLUMLEIN for HMVs with a square by the matrix number, and English Columbias with a ©, or in both cases with no code (post 1945) up to about 1953. BSI 78 for all post 1953 78s (in theory). HMVs with a diamond are American Victor recordings: use WESTREX.


You've already been pointed to this...
http://wiki.audacityteam.org/wiki/78rpm_playback_curves
...you can look up anything on there, and see how different it is from the "standard" curves. Unfortunately that table doesn't list the treble turn over frequency - quite an omission, because with it, you can make your own analogue circuit and/or create your own customer filter (in the equalizer software linked above - just type in the treble (=high) and bass (=medium) numbers, and the job is done! Few 78s use a LF roll off (=low) so you can leave that blank).

Hope this helps.

Very general rule (plenty of exceptions): the more modern the 78, the closer it is to RIAA, and the less it matters that you fix it.

Cheers,
David.

Cleaning up 78 transfers

Reply #42
a) trying to replicate what a completely analog phonograph horn sounds like is pointless.
Agree. It's easy to do, but there's more information and fidelity in these record grove than a gramophone/acustoustic phonograph can reproduce (though larger EMGs and HMV re-entrants do quite well).

However, within the limitations of the microphones of the time, the recording EQ was quite accurate. Especially early on, they didn't have many facilities to mess with it.

Cheers,
David.

Cleaning up 78 transfers

Reply #43
You should of course keep the original file (both before and after 'cleaning up' ... the 'before' in case you screw up something during the process),

I do a fair amount of photo editing, and really like the non-destructive side-car approach of e.g. Adobe Lightroom. You administer a database of untouched input files, and a matched database of editing recipies. If you change your mind about exposure, you simply move that slider, and everything else in the pipeline is repeated for the preview. If you want you can extract a subset of the edits and apply them to 100 images as a batch. As demosaicing processing improves, you can (at will) swap the process while still keeping your edits.

Is there something similar for audio? One would think that using VST-plugins or similar, you could get most jobs done inside one single application.

-k

Cleaning up 78 transfers

Reply #44
"I'm going to post this link for the third time. If you don't follow it this time, I'm giving up!"
"You've already been pointed to this..."

I see, 2bdecided... you're one of those guys who thinks that people who don't understand things are such a nuisance that you have to point it out to them publicly. Go ahead, give up. If you bothered to read anything I've said, I've made it pretty clear several times that I am looking for something that helps me understand how I would translate the numbers you keep throwing at me into usable curves that I can program myself into an audio editing program. You can throw all the charts and explanations about RIAA you want at me and remind me what I've already been pointed to but it doesn't explain in any way how to turn them into curves for labels that aren't listed. All those specs mean nothing to me if I don't know how to plot them. Does that make sense to you? Do you get what I'm not understanding? Spare me the hollier-than-thou nonsense about how inconvenienced you are by having to repeat yourself when you're not even reading what my questions are. You gave me some useful info but spare me the other crap.

Cleaning up 78 transfers

Reply #45
Quote
It's a bit to me like when a classic film is released with a new (and ridiculous) 5.1 mix when there was only ever center channel mono in the theaters of the time.
  Ha!  I've done that!    Well...  not with a classic film, but with a couple of rock concerts recorded in mono.  I did some EQ for a simulated-stereo effect, added some delay & reverb for rear channels, panned the talking-parts to the center, and panned the applause parts toward the rear.  Fun stuff!  Of course, the DVDs I made give the viewer a choice of original mono or synthesized 5.1 surround.

Cleaning up 78 transfers

Reply #46
That, at least, is adding to the atmosphere of a concert which was actually experienced by human ears. But those old films were recorded and mastered in mono and the sound was designed accordingly so everything would sound ok in theaters. It's like colorizing. Don't get me started...

Cleaning up 78 transfers

Reply #47
"I'm going to post this link for the third time. If you don't follow it this time, I'm giving up!"
"You've already been pointed to this..."

I see, 2bdecided... you're one of those guys who thinks that people who don't understand things are such a nuisance that you have to point it out to them publicly. Go ahead, give up. If you bothered to read anything I've said, I've made it pretty clear several times that I am looking for something that helps me understand how I would translate the numbers you keep throwing at me into usable curves that I can program myself into an audio editing program. You can throw all the charts and explanations about RIAA you want at me and remind me what I've already been pointed to but it doesn't explain in any way how to turn them into curves for labels that aren't listed. All those specs mean nothing to me if I don't know how to plot them. Does that make sense to you? Do you get what I'm not understanding? Spare me the hollier-than-thou nonsense about how inconvenienced you are by having to repeat yourself when you're not even reading what my questions are. You gave me some useful info but spare me the other crap.

Why would you need to program these curves into audacity (or any other audio editor) when the software I linked to applies them for you and saves the output as a new wav file? It even has an "RIAA recorded at 45rpm sped up to 78rpm" source preset. You couldn't want more, and it'll be 100x harder trying to do this just in audacity. Save the audio editor for the "by ear" EQ tweaks after you've got the basic reply curves in place.

I'm sorry if I offended you. I wasn't trying to humiliate you - I was just trying to get through to you that you've been given the complete answer four times now, and have ignored it.

Good luck.

David.

Cleaning up 78 transfers

Reply #48
Why would you need to program these curves into audacity (or any other audio editor) when the software I linked to applies them for you and saves the output as a new wav file?

I've got Capitol, Pathe, Odeon, 4 kinds of Columbia, Okeh and severael more that don't appear anywhere in that software. Yes, I did actually install it. What do I do about those? How do I begin to know which one to choose. That's all I was asking. If I get an unrelated answer 3 times, I'll restate my question 3 times. If you don't know anything about CDs and you want to know how to hear one song over and over again and all I tell you is "set it to repeat" that's completely useless because you don't even know what "repeat" is, much less where to find it. Do you understand what I'm saying?

Thanks for the apology, sorry to snap like that but nothing is more infuriating on boards than people who sound like they wanna remind others how much they know about something. More often than not, you're not likely to get an apology. So, acknowledged, accepted, and forgotten. As long as you can forgive my reaction as well.

Cleaning up 78 transfers

Reply #49
I've got Capitol, Pathe, Odeon, 4 kinds of Columbia, Okeh and severael more that don't appear anywhere in that software. Yes, I did actually install it. What do I do about those? How do I begin to know which one to choose.
If they are on that long list, and if the list included both turnover frequencies (which it doesn't), you'd just type those two values in the high and middle values of a custom filter. Where LF shelving is specified, type that in as the lower number. Where it's not specified just use 20 (or 1). Job done.

Unfortunately, one of the turnover frequencies is missing from all entries in that list! So, I'd look at the values for the type you have, and pick whichever standard curve is closest.

There must be a simple calculation to figure out the exact correct turnover value from the 10kHz cut value, but that's beyond me. It's something like...
treble_turnover_frequency = 5.6 - (0.25 * cut_at_10kHz)
...but that's a pocket calculator guess, not a proper formula.


I can give you MATLAB code to generate the curves from the turnover values though. I've just found the formula in some old EE notes. It's the a=... part below...

Code: [Select]
% riaa2.m - plot amplitude against frequency of RIAA filter

f1=50; % lower bass turnover / "low frequency" (if none use 1)
f2=500; % bass turnover  / "middle frequency"
f3=2121; % treble turnover / "high frequency"

f=1:20000; % set up array with frequencies 1-20k to calculate response at each frequency

% calculate response
a=(1+i.*f./f2)./((1+i.*f./f1).*(1+i.*f./f3));

a=a./abs(a(1000));

% uncomment this line to calculate inverse curve...
%a=1./a;

% plot amplitude of result in dB (using 20*log10) on log frequency scale (using semilogx) from 20-20kHz
semilogx(f,20*log10(abs(a)))
xlabel('frequency / Hz')
ylabel('amplitude / dB')
axis([20 20000 -30 30])


...but you don't need it. Unless you really want to program the curve into audacity.



Be prepared for some records not to match the listed curve at all. Sometimes labels licensed recordings from other labels, or went to different studios for one session. In this case, the named curve for that label could be completely wrong for that recording. If it sounds completely wrong to you, have confidence in your ears, and try another. At this point, it really is educated guess work. Trying increasing or decreasing the turnover frequency until it sounds right.

Hope this helps.

Cheers,
David.
P.S. forgotten.