Looking for help EQing an old recording Options

[MedO]

I was recently asked to copy an old tape to PC, which was copied from some other tape, and that might not have been the original medium either. Some asking and looking suggested that there is no easy way to get at a better version of the recording. The lyrics for many of the songs can't be found on the 'net and even for two songs where I found the lyrics and title, google did not turn up any page which contains both titles together.

So, this seems to be rare enough to warrant making a copy, and so I want to do a proper job. I already did some volume adjustment, applied mild noise reduction and restored missing parts of one channel with audio from the other. However, the result sounds like it could use some EQ.

However, since I'm not very experienced with audio editing, I'd like to ask if someone can give me advice on which settings to use, or perhaps even send me an Audacity EQ preset. I tried finding good settings myself, but didn't manage to judge whether changes I made were improvements or not.

I uploaded an excerpt of the audio here: http://www.ganggarrison.com/filehost/s1_excerpt.flac
Any help with the EQing or other suggestions would be much appreciated.

[MedO]

Hmm, I guess that would fit better in General Audio. Sorry for sorting it wrong, can someone move it?

[Notat]

Nothing obviously wrong with that. It would help if we could hear the unprocessed recording. I do hear some artifacts probably from the noise reduction.

It sounds like you started with a low bandwidth recording. EQ cannot bring back portions of the audio spectrum that were never recorded or that have been lost. If you aggressively EQ a noisy recording, you may get more balanced program but you often end up making the noise unacceptably prominent.

Trade offs everywhere. That's why mastering is more art than science.

[MedO]

Nothing obviously wrong with that. It would help if we could hear the unprocessed recording. I do hear some artifacts probably from the noise reduction.

It sounds like you started with a low bandwidth recording. EQ cannot bring back portions of the audio spectrum that were never recorded or that have been lost. If you aggressively EQ a noisy recording, you may get more balanced program but you often end up making the noise unacceptably prominent.

Trade offs everywhere. That's why mastering is more art than science.

I re-recorded the exerpt from tape (didn't keep the unedited version around because I was happy with the NR). Here it is. Sorry for the difference in volume
http://www.ganggarrison.com/filehost/s1_ex...t_unedited.flac

[Notat]

I'm hearing the same thing there plus noise. Might be tape speed flutter. There are some experimental techniques that can remove this. I'd take your best cut at it right now. Make sure to save an unprocessed version of the recording so that you can try again in the future when better tools are available.

[MedO]

I'm hearing the same thing there plus noise. Might be tape speed flutter. There are some experimental techniques that can remove this. I'd take your best cut at it right now. Make sure to save an unprocessed version of the recording so that you can try again in the future when better tools are available.

Heh, I actually thought about implementing something like that myself once, when I found a radio recording which had severe wow and still had the 19kHz FM stereo pilot tone.

Anyway, I don't think this problem can be solved with the technique described, since I don't see a clear marker frequency from the original recording here, and the audible 50Hz noise that I first thought was a candidate comes from my own tape deck (as well as quite a bit of the white noise). This is pretty strong since the recording level of the tape seems to be quite low. Besides, I didn't want to do a "perfect restoration" - it's already quite good as it is for my purposes, but I thought some EQing could make it sound a bit less tinny.

[markanini]

I'm hearing the same thing there plus noise. Might be tape speed flutter. There are some experimental techniques that can remove this. I'd take your best cut at it right now. Make sure to save an unprocessed version of the recording so that you can try again in the future when better tools are available.

I've considered these techniques myself for problematic tapes but there doesn't seem to be any readily available tools. I added a feature request to SoX 2 years ago and there hasnt been any activity yet. Also while in some cases a 15kHz FM pilot tone can make for ideal condition to carry out such processing capturing the tape bias at 40-50kHz could be difficult even with 96kHz sampling rate due to tape deck circuitry. Modifying the tape deck for half speed play back might allow for the tape bias to be captured in such cases(Or course then EQ will need correcting). FWIW

@MedO: Look up 'Tape Restore Live'.

This post has been edited by markanini: Oct 16 2011, 14:36

[markanini]

I spoke to early: http://www.celemony.com/cms/index.php?id=capstan
Price if crazy high though:€3,790

[MedO]

I spoke to early: http://www.celemony.com/cms/index.php?id=capstan
Price if crazy high though:€3,790

Just picking this up again; I played around with Tape Restore Live a bit and actually made a full run post-processing one side of the tape, but listening again now the result does not sound good to me (playing around with parameters on filters gets me tired so after a while I can't tell anymore whether I'm actually improving things or making them worse )

Anyway, I did play around with EQing before this attempt and ended up with this, which sounds better to me. Curious whether you agree.

http://www.ganggarrison.com/filehost/s1_excerpt_eq.ogg
Once more for comparison, un-eq'd version is http://www.ganggarrison.com/filehost/s1_excerpt.flac

[markanini]

These are the settings I would use for your sample with TRL:
This will only adjust high frequecies between channels to match per tape-to-head misalignment, though ideally this would have been done by adjusting the head alignment before transfer.

As for EQ that will have to be your decision, though I did play a bit with EQ on it. I saw an immediate improvement by cutting a narrow band around 4k, boosting 8k and 60Hz. From there on it's season to taste or leave it be. Boosting highs quickly adds noise so it's a balancing act. Here's the curve I arrived at for your reference:


What are you listening through. MedO?

[hlloyge]

Well, the noise is worse. When I was "restoring" my tapes, I would deal with the noise first, leaving recording as intact as I could.

[Arnold B. Kruege...]

I spoke to early: http://www.celemony.com/cms/index.php?id=capstan
Price if crazy high though:€3,790

Just picking this up again; I played around with Tape Restore Live a bit and actually made a full run post-processing one side of the tape, but listening again now the result does not sound good to me (playing around with parameters on filters gets me tired so after a while I can't tell anymore whether I'm actually improving things or making them worse )

Anyway, I did play around with EQing before this attempt and ended up with this, which sounds better to me. Curious whether you agree.

http://www.ganggarrison.com/filehost/s1_excerpt_eq.ogg
Once more for comparison, un-eq'd version is http://www.ganggarrison.com/filehost/s1_excerpt.flac

Not bad. ;-)

[krabapple]

(didn't keep the unedited version around because I was happy with the NR).

It's a good idea to always keep a 'flat' (unedited) transfer around, in case you change your mind.

[MedO]

Thanks, I will try your TRL settings. I wonder what boosting the 60Hz range will add except boosting the AC hum from the transfer, but will listen first and argue later . On the test file I uploaded before I reduced the frequencies around 1kHz and increased gradually after that.

This will only adjust high frequecies between channels to match per tape-to-head misalignment, though ideally this would have been done by adjusting the head alignment before transfer.

I did adjust the tape head, but only by ear (basically went for the position where the high frequencies were loudest).

What are you listening through. MedO?

Sennheiser PX-100s.

Well, the noise is worse. When I was "restoring" my tapes, I would deal with the noise first, leaving recording as intact as I could.

This is probably because the highs are boosted. I'm ok with having some noise left though.

[markanini]

PX-100 aren't ideal for doing EQ but you can use xnor's crossfeed to get a pretty good simulation of a stereo peaker set-up(affects EQ) and cross-check with your Hifi speakers. The 60Hz boost revealed a double bass that was inaudible before I liked the sense of space in the recording accentuated by boosting 8k but if you rather want more intelligible results you could try a smaller, narrower boost at 5-6k instead. I'd consider a narrow cut a 4k compulsory as it's a probem range in this recording, doesn't have to be wide, at least half an octave. I tried a cut at 1KHz as well, it really only worked well, to me, by using a super wide Q. You get more control by adjusting more ranges.
Anyway, you see why it's important to make your own EQ descisions, it really depends on which elements you want should stand out.

BTW, krabapple is right, you should keep the flat transfer.

EDIT: I forgot to mention I did'nt look at your noise reduction example. Personally I wouldn't bother with NR on this one.

This post has been edited by markanini: Nov 9 2011, 01:49