Bowie Station : View Topic: My guide to cleaning up noisy recordings (+ some samples)

Navigation

Poll

You must be logged in to view polls

Technical Help > My guide to cleaning up noisy recordings (+ some samples)

Welcome to our Forums	FAQ Search
	Controls · View New Posts · Mark All Read

You are in: Forums / Technical Help / My guide to cleaning up noisy recordings (+ some samples)

Fezziiwig

Posted at 2017-04-19 01:14:55(365 wks ago) (Technical Help / My guide to cleaning up noisy recordings (+ some samples))

Uploaded: 14.66 GB
Downloaded: 71.84 GB
Posts: 6

Ratio: 0.20
Location: United Kingdom

Greetings BowieStation,

As you can see, I'm new here and I've been greedily downloading bootlegs without passing on the favour, so I thought I'd at least try to make up for it by sharing some of the methods and tricks I've devised and discovered these past few days in my attempts to clean up noisy or otherwise damaged Bowie bootleg recordings. There are probably a million different discussions, books and published papers on the topic of digital audio restoration, but seeing as this has just been a little side project of mine to distract me from all the uni-coursework and exam revision that I should be doing, I haven't read into the topic in much detail.

I'm particularly keen on restoring 1976 and 1974 recordings because that period of Bowie is a favorite of mine. Here is a sample of what I've been able to do so far:

- From The Hunky Geordie Tapes, Vol 5 1976-05-05 Empire Pool, Wembley, London

TVC15 intro:
BEFORE -> AFTER

Life on Mars:
BEFORE
->
AFTER

Nothing miraculous but perhaps an improvement. Some bleepy artifacting can be heard in my cleaned versions, especially in the quieter parts but I've been getting better at avoiding that with practice.

Tip: Before you do anything, remove any DC offset that might be present in the audio.

The primary method I've used to reduce noise is by exploiting the differences between the left and right channels. The original recordings would have been done in mono (with one microphone) and put onto stereo magnetic tape so any differences between the two channels must be unwanted noise from either the tape or the circuits that recorded and reproduced the signal. The caveat is that in the real world there will be slight variations between circuits so the signals will also vary slightly

The crucial variation that occurs between channel's signals is their phase offset. If you analyse them, they are normally only 2 or 3 samples ( a few microseconds!) out of sync with each other but unless this is corrected, you cannot effectively isolate and remove any stereo noise. Once you've done that, it is a case of balancing the volume and dynamics of the various frequency bands of the signals. Do be aware, these can change over time. The frequency spectra, spectrogram and RMS contraster in audacity (which is the tool I use for all of this) are useful for making these adjustments. If anyone knows of any better tools for this, let me know.

When you have matched the channels to the best of your ability you can use a mid-side decoder to separate the similarities(mid) from the differences(side) of the two channels. If you've accurately matched the signals, the mid will be David himself with all the side-noise removed!

Alas, any noise that is common to both channels will still be standing between your ears and the words as they leave David's lips. There is often a low hum present in the sounds from the mains AC which appears at either 50 or 60Hz (and all the harmonics above and below it) depending on which side of the pond David refused to fly to when the recording was made.

There is no microphone (including the human (or animal) ear) which can perfectly convert a sound signal into an electric one however we can use equalisation to attempt to more accurately recreate the source sound. And we can use silent sections of the recordings to get the noise profile of the mic in order to reduce it that noise. It can be tricky to find silent moments in concert recordings because people applaud in-between the songs. Don't they know I'm trying to isolate noise? How inconsiderate of them.

At the end of the day, everything on earth is going to be subject to the RF interference from either broadcast towers, Wifi networks or the cosmic rays from stars, black-holes and the birth of the universe so you can never remove everything. Besides, my grandma seems to think that all this rock 'n' roll music is just noise anyway. She lived through the war and remembers the coronation, so she should know.

I hope at least someone will find this useful or at least of interest. I'm happy to try and answer any questions you might have, but I don't have any academic background in this field so my knowledge is limited. I could be completely wrong about everything that I've written. If anyone has any real experience or expertise with this sort of thing I'd love to hear anything you'd have to say.

Last edited by Fezziiwig on 2017-04-19 01:19:52

Steveboy	Posted at 2017-04-19 10:36:00(365 wks ago) (Technical Help / My guide to cleaning up noisy recordings (+ some samples))
Uploaded: 4.17 TB Downloaded: 1.08 TB Posts: 992 Ratio: 3.87 Location: United Kingdom	Hi Fezz Interesting but not sure the links are working Steveboy NO DISCUSSION I JUST WANT TO LISTEN TO THE MUSIC WANTED FRANKFURT 1976 1978 REHEARSALS

Fezziiwig	Posted at 2017-04-19 11:18:41(365 wks ago) (Technical Help / My guide to cleaning up noisy recordings (+ some samples))
Uploaded: 14.66 GB Downloaded: 71.84 GB Posts: 6 Ratio: 0.20 Location: United Kingdom	Hi Steve, Is there a way I can upload the files to BowieStation? What would be the best way to share the files seeing as they're just small mp3s? Fezziiwig.

Jean-Henri	Posted at 2017-04-19 13:08:11(365 wks ago) (Technical Help / My guide to cleaning up noisy recordings (+ some samples))
Uploaded: 97.90 GB Downloaded: 192.58 GB Posts: 243 Ratio: 0.51 Location: Netherlands	The links work for me. Life on Earth?

Steveboy	Posted at 2017-04-19 18:52:15(365 wks ago) (Technical Help / My guide to cleaning up noisy recordings (+ some samples))
Uploaded: 4.17 TB Downloaded: 1.08 TB Posts: 992 Ratio: 3.87 Location: United Kingdom	Been to work and now back and the links now work..... Why dont you upload the complete show as a torrent?? Steveboy NO DISCUSSION I JUST WANT TO LISTEN TO THE MUSIC WANTED FRANKFURT 1976 1978 REHEARSALS

Fezziiwig

Posted at 2017-04-19 23:35:08(365 wks ago) (Technical Help / My guide to cleaning up noisy recordings (+ some samples))

Uploaded: 14.66 GB
Downloaded: 71.84 GB
Posts: 6

Ratio: 0.20
Location: United Kingdom

Quote:

Why don't you upload the complete show as a torrent??

I'm just practicing at the moment, so I've just been working with individual songs. I'd like to get a little better at it before I attempt a full show. When I do, if people would be interested, I'd be more than happy to submit it as a torrent.

40compasses	Posted at 2017-04-21 07:43:24(364 wks ago) (Technical Help / My guide to cleaning up noisy recordings (+ some samples))
Uploaded: 118.42 GB Downloaded: 63.69 GB Posts: 41 Ratio: 1.86 Location: unknown	More power to you, Sir! I too would definitely like to hear more clarity in some of the shows of your favoured period.

abogner90	Posted at 2017-04-22 12:27:27(364 wks ago) (Technical Help / My guide to cleaning up noisy recordings (+ some samples))
Uploaded: 2.42 GB Downloaded: 17.11 GB Posts: 85 Ratio: 0.14 Location: United States of America	Fezziiwig check out the german company,brainworx. They make really good audio plugins to clean up audio. There are also some free plugins to get you started, i.e. M/S technology. It could make a big difference in your 'clean up' duties. Andy

billy1

Posted at 2017-04-23 17:45:44(364 wks ago) (Technical Help / My guide to cleaning up noisy recordings (+ some samples))

Uploaded: 108.28 GB
Downloaded: 28.82 GB
Posts: 2

Ratio: 3.76
Location: United Kingdom

The problems with mono recordings as I see them. The original recordings have two channels with some differences, they are never the same due to the recording head azimuth which no matter how well it has been set will never give you two equal channels. This can be exacerbated when the original tape is copied onto another tape. Now some people like to keep both channels if, say, the left has better bass response, but the right has better highs and put up with any 'stereo' effect from the two different mono channels. In the case of the samples you provided the eq differences are negligible and so you should drop one of the channels, say the one with the least number of clicks, drop outs and then do any work on that single channel, then duplicate it when you're ready to prepare your tracks for burning to CD.

neilwilkes

Posted at 2017-04-24 10:07:54(364 wks ago) (Technical Help / My guide to cleaning up noisy recordings (+ some samples))

Uploaded: 761.82 GB
Downloaded: 627.05 GB
Posts: 724

Ratio: 1.21
Location: United Kingdom

Mono recordings often have a very limited sense of depth as well, making them sound un-natural.
This can be got around in great part though by careful use of impulse response plugins such as AltiVerb or Waves IR-1 and IR-360 which can be used to make the source file sound as if it were recorded in any of the famous venues & studios are modelled. This can work really, really well & has even been used on a version of the Desk Tape from the final Ziggy show - cannot remember who did it off the top of my head but he did a neat job.

Some notes to the OP.

1 - Removing background & broadband noise should be done in bites if you want to reduce the artifacts.
Find a section with just the noise, print this in something like iZotope's RX-5 DeNoiser & take out around 3dB of the noise level and not the default 12dB of reduction. Reprint it & do this again.
You need to be careful as this process will both expose any other encoding artifacts such as the tell-tale chirping or space monkey sound of lossy encoders or the muffled results of losing too much in the top end.

2 - Mains Hum should be removed before any broadband NR is done - all you need is to load the default for the territory & then "learn" - again the RX-5 DeHum tool zeroes in on the actual hum frequency as our mains may be stated to be 50 or 60Hz but the reality can be a few Hz upwards in both cases - I have seen a fundamental as high as 75Hz which really did shock me (although I may have done a speed correction beforehand - I cannot remember, but it does seem the more likely explanation in reflection)
But this noise can - and should - be eliminated.

Finally, can I suggest that whenever you offer a denoised product - or any alternate version that has had any form of processing carried out on it - to also please include the original version? Tools get better - as will your experience in using them, and I speak from my own experience as even though I do this for a living I still cringe at the sound & sight of some of the things I was doing 15 years ago now as not only have I got better at doing this stuff but the tools themselves have got a lot better too.

billy1

Posted at 2017-04-24 16:05:23(364 wks ago) (Technical Help / My guide to cleaning up noisy recordings (+ some samples))

Uploaded: 108.28 GB
Downloaded: 28.82 GB
Posts: 2

Ratio: 3.76
Location: United Kingdom

[quote=neilwilkes] Mono recordings often have a very limited sense of depth as well, making them sound un-natural.This can be got around in great part though by careful use of impulse response plugins such as AltiVerb or Waves IR-1 and IR-360 which can be used to make the source file sound as if it were recorded in any of the famous venues & studios are modelled. This can work really, really well & has even been used on a version of the Desk Tape from the final Ziggy show - cannot remember who did it off the top of my head but he did a neat job.Some notes to the OP.1 - Removing background & broadband noise should be done in bites if you want to reduce the artifacts.Find a section with just the noise, print this in something like iZotope's RX-5 DeNoiser & take out around 3dB of the noise level and not the default 12dB of reduction. Reprint it & do this again.You need to be careful as this process will both expose any other encoding artifacts such as the tell-tale chirping or space monkey sound of lossy encoders or the muffled results of losing too much in the top end.2 - Mains Hum should be removed before any broadband NR is done - all you need is to load the default for the territory & then "learn" - again the RX-5 DeHum tool zeroes in on the actual hum frequency as our mains may be stated to be 50 or 60Hz but the reality can be a few Hz upwards in both cases - I have seen a fundamental as high as 75Hz which really did shock me (although I may have done a speed correction beforehand - I cannot remember, but it does seem the more likely explanation in reflection)But this noise can - and should - be eliminated.Finally, can I suggest that whenever you offer a denoised product - or any alternate version that has had any form of processing carried out on it - to also please include the original version? Tools get better - as will your experience in using them, and I speak from my own experience as even though I do this for a living I still cringe at the sound & sight of some of the things I was doing 15 years ago now as not only have I got better at doing this stuff but the tools themselves have got a lot better too. [/quote]

This is more like it. Thanks Neil. Yeah, some people hate listening to mono recordings - I personally find them OK on my main speakers but not so good on headphones. My advice is do what you want, either pure mono or with a widening or depth effect - if anyone complains at either decision just give them a link to what is (hopefully) the now inferior source. One things for sure, if you have improved the eq and have things at the correct pitch very few will moan about either decision.
Here's more from me, with the caveat that I have no qualifications in this and am just sharing my own experience.

PITCH CORRECTION

To me pitch correction really should be your first move - if you don't know how to do it think about asking someone else to do it for you.

Really, pitch correction requires knowledge of musical keys generally, the keys used live, sometimes they differ and a reference containing the frequency values of individual notes. Oh, and a frequency analyser like the one in Adobe Audition.

A guitar, keyboard or virtual keyboard at hand is also a must in my opinion.

If their is piano in the music you should be checking and adjusting the pitch with reference to those piano notes, if not then organ next. If not that, then guitar. Avoid bass if you can. In all cases you should be checking mid notes - a piano tuner will stretch the upper and lower part of a keyboard which means at some point the highest notes become progressively sharper and the lowest notes progressively flatter. It's to do with harmonics, how one note sounds against another. Best to stick to the middle of the keyboard - same with guitar due to possible intonation issues. You will find that it's not possible to get A=440hz, 880hz, 1760hz etc, not exactly, you can get one bang on, but the others may be slightly off.

In my experience around 0.3% sharp or flat is an acceptable result, if you get that it will sound spot on to just about anybody - and anyone with a guitar or keyboard will be able to play along.

Sooner rather than later you will find that the pitch correction needed at the end of a piece of music is greater or lesser than it was at the start. This is caused usually by failing batteries in the source machine or the machine used to make a copy. Here you need to perform what Adobe Audition calls a gliding stretch. Instead of one percentage value for the stretch/adjustment you use two. If you subsequently find the music in the middle still sounds off you can go back and try pitch correcting in 30 sec or 1 min sections - or whatever you feel is needed.

The advantage of doing it in shorter segments is you're more likely to find any places where the pitch suddenly drops or rises. If you think about it, a gliding stretch over such a section would be a mistake. In fact you should actually listen right through the tape to spot any instances.
Warning! As you can imagine, this can all be very time consuming.

EQ

The most common mistake people make when eq'ing an old tape is boosting the mids/ upper mids - presumably to bring up David's voice or Ronno's guitar. That's completely the wrong approach and in almost every case is the exact opposite of what you should be doing.

Many of these old tapes have a frequency response that is kinda like a hump - they favour the mids and don't do as well in the highs and low end.

The human voice and guitar contain more frequencies than most people realise. These frequencies are already out of balance due to the mic and tape used.

On a two track stereo or mono tape you can't isolate the voice or guitar to boost it. They share frequencies with all the other instruments. What usually happens is that the frequencies that already stand out due to the aforementioned mic/tape limitations get boosted further to the point where they cause harsness or even distortion.

What needs to be done is to bring all the frequencies back into a more natural sounding balance. This means reducing the gain on some frequencies, not boosting.

How to isolate those frequencies?

Well, the method I favour involves the parametric equaliser in Adobe Audition. What you do is create a point by selecting one of the numbered boxes on the left, give it a narrowish Q/ Width, say 1. Now raise the volume way up and sweep the point left to right/, right to left - it will be obvious when you come across what we will call the resonant frequencies (don't do this through your main speakers). You may find around half a dozen - I find 4 or 5 is enough. One of them will be what is causing mud in the bass area (it was a demud preset that gave me this idea in the first place). Now move all these point below 0db say -3db and give them a larger Q/ Width. Play around with the values and audition your changes via the power button at bottom left.

This is partly how I remastered the Ziggy show I upped a month or so ago, though once I had isolated my frequencies I actually used a dynamic eq plug in to reduce them.

Don't do it all through headphones or your PC speakers. Now and then hook it up to see how it sounds through your main Hi Fi speakers.

Fezziiwig

Posted at 2017-04-24 16:26:24(364 wks ago) (Technical Help / My guide to cleaning up noisy recordings (+ some samples))

Uploaded: 14.66 GB
Downloaded: 71.84 GB
Posts: 6

Ratio: 0.20
Location: United Kingdom

Thanks for your excellent and supportive advice guys!

Quote:

Finally, can I suggest that whenever you offer a denoised product - or any alternate version that has had any form of processing carried out on it - to also please include the original version?

I actually did include the original versions along with my samples, but you make a very good point. I'm very busy with uni deadlines and exams at the moment so I'm on a little hiatus from learning and practicing all of this stuff but I can't wait to get back into it in a couple weeks.

Besides free time I'm also severely limited by money which is why I can only rely on free tools like audacity. The amount of free plugins available online actually make it very flexible and I'm learning to write my own plugins and programs for it too.

<< Prev 1 Next >>

Locked

You are not permitted to post in this forum.

Latest Forum Posts

Latest Topic Title	Replies	Views	Author	Last Post
Lady Grinning Soul	22	1457	gique	by Besomdave 2024-04-18 18:47:47
"Changes" and "TVC15" on "Live Nassau Coliseum '76" album	38	4511	yasuko	by russellmael 2024-04-18 15:52:10
RIP Dennis P. McCann (Paperdragon)	11	235	JanErik	by Davesolar 2024-04-18 11:13:51
What is the worst bootleg from each david bowie tour	34	5386	averybowie	by Stranger09 2024-04-17 23:27:43
Ziggy Box Track List	26	1873	Steveboy	by Steveboy 2024-04-10 18:15:20
Qbittorrent Stalled	14	2128	homebrew	by Portlandbill 2024-04-10 16:11:35
RSD 2024 - Underwhelming or what?	20	1373	neilwilkes	by professormouse 2024-04-08 23:45:01
So what is left ??	24	1819	Portlandbill	by professormouse 2024-04-08 23:38:23
DB in 360 Spatial Audio	4	352	marksach	by Steveboy 2024-04-06 16:03:44
Rainbowman - Jerome Soligny	2	440	SteveMaturin	by neilwilkes 2024-04-02 11:47:30
Imagine the Possibilities	10	787	homebrew	by professormouse 2024-03-18 00:40:04
Chilli Pepper 1997 - New Vinyl Release, but.....	26	2226	Steveboy	by Portlandbill 2024-03-09 22:43:14

Online Users

Newest Members