Corrrect processing order for 78s recorded at 45

[waxcylinder]

Ken, thanks for the update

WC

[waxcylinder]

I have finally gotten around to creating the promised tutorial in the Audacity manual under development for 1.3/2.0 on the recording and processing of 78s with Audacity. See: http://manual.audacityteam.org/index.ph ... pm_records

The information in this tutorial was garnered and condensed from several threads on this topic in the forum and based on an original short article in the Wiki (provided by Gale Andrews, I believe). So my thanks go out to all the forum contributors who have provided insights and experience for this tutorial.

Those who should be mentioned in despatches for their inputs to this tutorial (in no particular order) are:
pdxrunner
L_Libza
bobv
scar
BillW
SteveTF
Koz
whomper

Thanks,
WC

[bobv]

@waxcylinder - Nice job on the tutorial.

I finally got back to my project today and started to do some cleanup work. I found a possible problem with doing the normalize step before removing clicks. The clicks are very loud, so normalizing to -1 db lowers the track's volume. If I get rid of the clicks first, normalizing raises the volume.

Is there a reason to normalize before high/low pass filters or noise removal are applied? Or should normalization be delayed until after other filtering/click removal are done?

[whomper]

@waxcylinder - Nice job on the tutorial.

I finally got back to my project today and started to do some cleanup work. I found a possible problem with doing the normalize step before removing clicks. The clicks are very loud, so normalizing to -1 db lowers the track's volume. If I get rid of the clicks first, normalizing raises the volume.

Is there a reason to normalize before high/low pass filters or noise removal are applied? Or should normalization be delayed until after other filtering/click removal are done?

if noise clicks are high
they limit the amount of gain for normalization

i would normalise next to last
compress last
do all the other stuff earlier

[waxcylinder]

Many thanks for the feedback bobv and whomper: yes Icertainly when I record my vinyl I do capture, then click removal is my very first processing step (I use Brian Davies' ClickRepair tool). And I do amplitude adjustment as my very last step prior to export. For stereo vinyl I use the Amplify effect rather than Normalize - as Audacit'y's Normalize works on each stereo track independently, so can alter the stereo balance. But for mono 78s I am happy to recommend the Normalize for their amplitude adjustment.

Looking at the tutorial again, I realize why SteveTF originally included the Normalize step at that stage: it was primarily to remove any DC offset that may be in the recording. IMHO it's rather odd that Audacity is currently parceled in with the Normalize effect (there are discussions on Feature Requests in the Wiki and on the developer mailing lists to change this - but don't hold your breath).

So In the next day or so I will tweak the order of the workflow - DC offset removal and click removal early - and Normalization late in the process. An include an optional compression (I did have that on my original draft of the tutorial - but for soem reason I neglected to copy it over).

WC

[steve]

It might be of interest - the current Audacity source code includes support for "Inverse" Equalization curves (great for doing the "reverse RIAA" thing). This may be included in the new Audacity 1.3.12 release that is due next week.

[L_Libza]

Good start on the tutorial, Waxcylinder.

Just a couple of additional thoughts on the transfer of 78s from my own personal experience. I have done hundreds of transfers and I feel my technique has improved with experience. In fact I have redone some of my earlier work because of software advancement and a better understanding of the process.

If your turntable doesn't have a strobe or speed adjustment, you can measure the RPMs very accurately by recording the playout of the end groove and selecting the distance between the recorded clicks on the wave form on the waveform display. I measure the time for 10 consecutive revolutions which I can actually measure to 1/1000 on a second by zooming in on the waveform. Then it is simple mathematics to determine the actual rpm. I have made a spreadsheet which gives the appropriate speed correction based on the measured elapsed time.

I had the problem of DC offset which meant I had to run a "normalize" every time. I solved this by changing to a USB audio interface which has totally eliminated the DC offset. I use a Behringer UFO202 with which I am very satisfied.

Don't expect miracles with badly worn records. The process can be very frustrating and the results can be disappointing. Avoid aggressive denoise. The artifacts are usually worse than the noise. I like to leave a little surface noise in my transfers (they are 78s after all!). Declick and equalization are the most important steps in the process. Learn to read the waveform. Sometimes an equalization can increase the amplitude of some frequencies to a clipping level, so consider reducing the amplitude slightly before equalization.

[waxcylinder]

... I like to leave a little surface noise in my transfers (they are 78s after all!). ...

You're just like my son.

When I made transciptions of the 45s that had lived on my jukebox for years (and had a very hard life there) I cleaned them all up nicely. He complained "that's not the way they're supposed to sound dad" - so I had to make him a copy of the unprocessed capture recordings. "Much better" he said - "just like they sounded on the juke box".

And thanks for the further feedback - much appreciated.

WC

[waxcylinder]

Ok - I updated the tutorial this morning following recent feedback: adjusted the workflow order and added a few notes.

See: http://manual.audacityteam.org/index.ph ... pm_records

Thanks,

WC