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 55572 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Cleaning up 78 transfers

Reply #50
(Mal)apropos: is there is any reason for applying EQ before cleaning up pops? (Edit: same goes for speed adjustment.)

Otherwise, I would suppose this would be a Great Idea [tm] for Mudlord to improve the http://www.foobar2000.org/components/view/foo_dsp_effect component . (Do it tag-based!) Edit: Or maybe for foo_convolve.

Cleaning up 78 transfers

Reply #51
So, I'd look at the values for the type you have, and pick whichever standard curve is closest.

I really appreciate the effort to get me to understand this but I'm sorry it's just way too over my head.

Here, for example, are the numbers on that chart for Capitol:
Bass Turnover Frequency (Hz): 400
10 kHz Gain Rolloff (db): -12
Source: 3
And Equalizer gives me the option to enter numbers for High, Medium and Low Frequency. Can you see why I'm confused?
Questions:
- Is the bass turnover frequency the High, Medium, or low?
- How can I use 20 or 1 for LF shelving? Equalizer doesn't even let me enter a value for shelving. Also, I don't understand what these are referring to so I don't get why a variation of 20 - 1 isn't a big jump.
- How do I compare what I have to the curves present if they deal in different values?

I think I'm personally better off capturing the audio, inverting the RIAA curve, cleaning clicks and pops, and then speeding the 45rpm up to 78 and just adjusting by ear. I just don't see any other way.

Cleaning up 78 transfers

Reply #52
On this page...
http://www.rfwilmut.clara.net/repro78/repro.html#eq
...compared with the labels in equaliser, high is called "Treble turnover", middle is called "Bass turnover", and low is called "Lower bass t/o".

That's at least easy, because it has all three numbers.

Where there's no Lower Bass turnover listed, using 1 or 20 is virtually the same because both 1Hz and 20Hz are below the audible range, so won't do anything.


Problem with this...
http://wiki.audacityteam.org/wiki/78rpm_playback_curves
...is that two out of three numbers are missing.

So, in your example...
Quote
Capitol:
Bass Turnover Frequency (Hz): 400
10 kHz Gain Rolloff (db): -12
Source: 3

Use 20 for Low (because it doesn't have one)
Use 400 for Medium (because that's what it says for Bass)
Use 2500 for High (because EMI LP has a -12dB cut at 10kHz, just like Capitol, and EMI LP has a treble turnover frequency of 2.5kz=2500Hz)
or use 2600 for High (because 5.6-(0.25*12)=2.6kHz=2600Hz - formula from my last post)

Or just use EMI (33) LP because it's almost the same numbers (see that long list). Or even RIAA because it's not that far out in this case (again, see numbers in list).

Hope this helps.

Cheers,
David.

Cleaning up 78 transfers

Reply #53
I just heard that whole post in the voice of Benedict Cumberbatch in Sherlock during one of the deductive reasoning sequences.
Brilliant. Thank you.

Now... one last question... 
Why is that these curves need to be applied? Here's my thinking, tell me where I'm wrong:
- The audio was recorded onto the original wax or shellac master with the thought it would be played on a phonograph through a reproducer and therefore needed a certain EQ in the studio
- Our contemporary machines not only do not know that EQ but also are adding RIAA's EQ
- The inverting of RIAA cancels out what our stereos do
Wouldn't that then leave the same sound that a phonograph reproducer and stylus would have picked up and sent through a phonograph horn?

Cleaning up 78 transfers

Reply #54
Yes, but:

- 30's and 40's recordings were -- I guess your mileage may vary here -- made for ampified reproduction, not for horn phonographs
- and in the phonographic horn era, fidelity was not that much of an issue: getting sound out was already a victory over nature. Just because they couldn't apply EQ without an amplifier, that doesn't mean the output was ideal.  Look at http://en.wikipedia.org/wiki/Gramophone_record#Equalization , quote: In 1926 Joseph P. Maxwell and Henry C. Harrison from Bell Telephone Laboratories disclosed that the recording pattern of the Western Electric "rubber line" magnetic disc cutter had a constant velocity characteristic. This meant that as frequency increased in the treble, recording amplitude decreased. Read also the http://en.wikipedia.org/wiki/Gramophone_re...#Sound_fidelity part.

Cleaning up 78 transfers

Reply #55
Ok, so with 2bdecided's help I'll be doing the following:
- RIAA curve on the original (45rpm) recording
- correct the speed to 78
- apply the correct curve
And I'll be doing my click repair and noise reduction right after applying the RIAA inverted curve (so after step 1 in this case) because I think it does a better job repairing at the slower 45.

Does that sound alright?

Cleaning up 78 transfers

Reply #56
Answering your previous point...

I don't know the full story, but I believe the basic 6dB per octave curve comes from the amplitude <> velocity conversion inherent in many pick-up types. If you double the frequency, but maintain the amplitude, the needle moves twice as far per second (because there are twice as many back/forth wiggles) = twice the signal from most pickups. If you double the frequency, and try to keep the needle movement amount the same, the amplitude will have to halve (twice as many wiggles + half the amplitude) = same amount of signal from most pickups.

So the basic response curve, falling 6dB per octave top left to bottom right on a graph, is due to this.


The various different kinks that are put into the middle of this graph by various choices of turnover frequencies are different attempts to twiddle this basic 6dB per octave curve to cope with various amounts of bass + treble that various recording companies wishes to put onto records while maintaining a reasonably close groove spacing / long recording time. There was also the need to avoid huge bass excursions (which cause needles, of any type, to jump).

There must have been some attempt to make records match gramophones, and radio-gramophones match records, from each company. Especially in the UK, all the largest companies made both.

Maybe someone knows more?

Cheers,
David.

Cleaning up 78 transfers

Reply #57
Doing click repair after step 1 also means that you will be processing unequalized data. Are you sure that is what you want?

It means that if you listen to the effect of click removal you will be listening to a version with greatly exaggerated high frequencies.

Cleaning up 78 transfers

Reply #58
I'd fix the speed first, then run it through equaliser to remove RIAA and apply correct EQ, then declick etc.

Try it other ways as a test if you want, but unless you have a weird declicker, the above is usually the easiest and best.

Cheers,
David.

Cleaning up 78 transfers

Reply #59
I'd fix the speed first, then run it through equaliser to remove RIAA and apply correct EQ, then declick etc.

Confused a bit... I was going by what you said earlier... "That's the wrong way around. You need to invert the RIAA curve on the original (45rpm) recording, correct the speed (to 78), and then apply whatever you think the correct curve is."

Cleaning up 78 transfers

Reply #60
I'd fix the speed first, then run it through equaliser to remove RIAA and apply correct EQ, then declick etc.

Confused a bit... I was going by what you said earlier... "That's the wrong way around. You need to invert the RIAA curve on the original (45rpm) recording, correct the speed (to 78), and then apply whatever you think the correct curve is."

You're right about this. RIAA inversion before speed adjustment.

Cleaning up 78 transfers

Reply #61
Doing click repair after step 1 also means that you will be processing unequalized data. Are you sure that is what you want?

My thinking with 78s is that doing click repair after RIAA but before anything else allows the software to run click detection at 45 and picking up things it might miss going at 78. Then once clean(er), I can speed it up and EQ if necessary. Is there any problem with that line of thinking? My only question is: does going at 78 and having the clicks be at a significantly higher frequency than the music make them easier to detect than if they're going so slow that they could be confused for being part of the music? Or maybe it doesn't work like that?

Cleaning up 78 transfers

Reply #62
The music and clicks would both be moved to higher or lower frequencies together. It should have no impact on the ability to distinguish between them, unless the software makes assumptions about what frequencies are audible.

Cleaning up 78 transfers

Reply #63
Ok so FINALLY cleaned everything up but unfortunately ended up with audio that's about 3db quieter. I guess that makes sense with all the cleaning. Is it ok to raise the levels a bit or does that do something awful to the file?

Cleaning up 78 transfers

Reply #64
Ok so FINALLY cleaned everything up but unfortunately ended up with audio that's about 3db quieter. I guess that makes sense with all the cleaning. Is it ok to raise the levels a bit or does that do something awful to the file?


I used to be a purist and not touch an audio level in digital. I now do it all the time similar in concept to replay gain but my level changes work everywhere, not just compatible players. If you're not comfortable with it, change the level and save with a new file name. My guess is you won't be able to tell the difference. BTW I've used CoolEdit and then Audition (same thing, new owner) exclusively since 1998 - currently Audition 3.0


Cleaning up 78 transfers

Reply #65
I'd fix the speed first, then run it through equaliser to remove RIAA and apply correct EQ, then declick etc.

Confused a bit... I was going by what you said earlier... "That's the wrong way around. You need to invert the RIAA curve on the original (45rpm) recording, correct the speed (to 78), and then apply whatever you think the correct curve is."

You're right about this. RIAA inversion before speed adjustment.
Only if the only thing you have available is the inverse RIAA curve

Equaliser helpfully has a "RIAA curve that was recorded at 45rpm then sped up to 78rpm" inverse curve, so this will work just fine.

Cheers,
David.

Cleaning up 78 transfers

Reply #66
Doing click repair after step 1 also means that you will be processing unequalized data. Are you sure that is what you want?

My thinking with 78s is that doing click repair after RIAA but before anything else allows the software to run click detection at 45 and picking up things it might miss going at 78. Then once clean(er), I can speed it up and EQ if necessary. Is there any problem with that line of thinking? My only question is: does going at 78 and having the clicks be at a significantly higher frequency than the music make them easier to detect than if they're going so slow that they could be confused for being part of the music? Or maybe it doesn't work like that?
It depends entirely on your declicker.

If a particular declicker worked better at half speed with the treble boosted, the declicker author could build that processing in, couldn't they? Not to say that they would. Some declickers do seem to need a bit (or lot) of help.

Some of the best delickers understand some psychoacoustics. That'll be partly broken by running at completely the wrong speed + EQ.

Ok so FINALLY cleaned everything up but unfortunately ended up with audio that's about 3db quieter. I guess that makes sense with all the cleaning. Is it ok to raise the levels a bit or does that do something awful to the file?
You can do what you like to the levels, as long as they don't clip. Peak normalising everything is fine, though you'll probably end up with different tracks sounding different loudness. I'd use ReplayGain, but then I would say that . Quickest thing is to run Wavegain across all the files, though this may still leave them all quieter than you want (though all roughly the same loudness).

Cheers,
David.

Cleaning up 78 transfers

Reply #67
Equaliser helpfully has a "RIAA curve that was recorded at 45rpm then sped up to 78rpm" inverse curve, so this will work just fine.


Good! Are there then any phase issues (or others) which would make it beneficial to actually apply only the difference EQ curve once?

(I'd bet it won't matter anything for even the highest-fi 1933 '78 ...)

Cleaning up 78 transfers

Reply #68
Equaliser claims to get the phase right too, though I haven't checked myself.

(I'm not plugging this software, honest - I hadn't heard of it myself until this thread!)

Cheers,
David.

Cleaning up 78 transfers

Reply #69
Ok, I'll increase levels on my own without clipping. Just want to get them peaking at 3db to avoid having to crank the speakers. But it sounds great.
Applied inverted RIAA, sped up to 78, applied the custom "Capitol EQ" 2bdecided helped me with, removed clicks and noise, sounds wonderful. A far jump from the beaten up original. It's just a little quiet.

(...still think the song sounds a bit better without the RIAA curve inverted though. Seems to lose too much bass. More than what sounds appropriate. I guess that's where the tweak by ear comes in. Maybe a bass boost.)

Cleaning up 78 transfers

Reply #70
Yes, I usually find the "correct" curve to be lacking in bass. Not always, but mostly. Some re-issues of 78s (especially some American ones from the 1970s/80s) boost the bass by a ridiculous amount.

It's quite hard to judge by ear though. Do the best you can, but keep the previous version in case you decide you can do better later (e.g. when you change stereo or headphones).

FWIW I find it really difficult because I'm so used to listening to my records with the RIAA curve applied. It's only through the PC that I have access to the correct curves. I can hear that the treble especially is more natural with it corrected, but it's not what I'm used to from my favourite records.

Cheers,
David.



Cleaning up 78 transfers

Reply #73
FWIW when I was doing my major "78 project" I was in touch with collectors all over the world. They all had several different size/shape styli for different labels/periodsbut invariably listened via RIAA eq. I guess that's partly because RIAA was effectively a default being built in to every amp that had disc replay capability. But knowing the lengths some of them went to in pursuit of their hobby I'm sure they would have spent whatever was necessary to get the right eq if they thought it mattered that much - or maybe they were more interested in the music than the technicalities as 2B says

Cleaning up 78 transfers

Reply #74
FWIW when I was doing my major "78 project" I was in touch with collectors all over the world. They all had several different size/shape styli for different labels/periodsbut invariably listened via RIAA eq. I guess that's partly because RIAA was effectively a default being built in to every amp that had disc replay capability. But knowing the lengths some of them went to in pursuit of their hobby I'm sure they would have spent whatever was necessary to get the right eq if they thought it mattered that much - or maybe they were more interested in the music than the technicalities as 2B says
You may have just known the wrong collectors. The two I know with many styli also have dedicated analogue pre-amps giving any possible curve.

Cheers,
David.