The Missing Octave(s) - Audacity Remastering to Restore Tracks

[Chris A]

I'm not sure what the issue is.

 

Absolutely no issue!   The sound starts in your head,  you can try to have it reproduced in your room, to match what you might expect it to be. 

 

You can not get there here.  Earth plane experience will always be lacking.

 

I'm not sure what you are saying here, but in the spirit of the forum, I'll take this is the best way possible. 

 

My experiences are always in the earth plane, some more enjoyable than others...    I hope that you're enjoying yours, too.

 

I'm enjoying this particular exercise, since it lays to rest so many issues that I've had in the past with music (and blamed hardware in the form of loudspeakers, room acoustics, and electronics) that I've owned and have been listening to for decades now.

 

I'd say "catharsis" is a better description.  YMMV.

 

Chris

Edited February 1, 2015 by Chris A

[Axz Hout]

g

[dubai2000]

It ain't easy......okay, what I noticed is that after hitting normalize the whole track is less loud (using headphones on my computer to check things)......What am I missing here?

 

Wolfram

[Chris A]

Make sure that the Normalize dialog text box has a "-0.3" in the maximum dB field, and not some figure that is more negative - such as "-10", etc.

 

If the track that you selected is a very loud one, when you hit Normalize... and hit OK, the whole track might be reduced as much as 0.3 dB, which isn't very much - almost imperceptibly.

[mustang guy]

It ain't easy......okay, what I noticed is that after hitting normalize the whole track is less loud (using headphones on my computer to check things)......What am I missing here?

 

Wolfram

Real quick, my understanding of 'Normalize' in Audacity.

 

When you normalize, you are setting the highest db in the song to a specific db from the maximum before clipping occurs. If you normalize to -1, then the max db in the song is made to be 1db under clipping, and the rest of the song is adjusted by the same amount. This could boost or de-boost a song. The thing it does is make headroom so you don't end up clipping when you EQ boost the low frequencies. 

 

In the one I did, I had to Normalize at -3.5db to keep it from clipping. After I did my adjustments, I used normalize and set the amount to -0.1, to make it as loud as possible without clipping. If could very well cause the overall song to be less loud, but the bass will now be much louder as compared to the rest of the song. It's a trade off, but one most audiophiles are very happy to make.

[mustang guy]

Billy Squire - Everybody Wants you:

 

 

I'm getting better at this. Note the clipping before and after. I allowed some clipping so I wouldn't use overall volume so much. I cannot tell one is actually louder than the other. In fact, the one on the right is fuller and therefore SOUNDS louder.  

 

Here is the Dynamic range data:

 

before adjustments:  Left 8.27 dB   ---      Right 7.96 dB 

after adjustments: Left 10.21 dB   ---      Right 9.87 dB

 

This is sort of an indication that I ended up -1.9db less loud in the song overall, but with louder at 60Hz down, plus a little adjustment at 100 Hz.

[mustang guy]

I just had a chance to test the Money For Nothing update I did on stereo LaScalas and dual THT subs, and the difference is VERY noticeable, and in a positive way. The unchanged version had no thump at all coming from the subs. The converted version took full use of them. This is most definitely a worthwhile project!

 

Chris,

 

For what it's worth, you have my stamp of approval. I am a believer.

[muel]

Couple of things... 

 

I assume that bumping up the bottom octave could also result in bumping up some distortion.  If I run into any unacceptable level that I perceive I can always lower the EQ a bit.  I would like to hear some comments about distortion (e.g. speakers working harder to produce what they can't).

 

This is a fun experiment for me!  It is completely harmless because if I don't like the results I can just delete the EQ'd versions I don't like.  Worst case I've wasted some time but I found myself listening to the EQ'd Diana Krall all the way through.  I often skip around so that says a lot about how I liked the results.  I will be playing with this some more for sure.  The bottom line is how much enjoyment I get out of the music and so far this looks to be a benefit.  

 

Thanks for presenting this to us here, Chris!  You have put in some work and anyone who wants to pursue can also benefit!

[Axz Hout]

j

[Chris A]

No worries - I realize that this is in part a leap of faith effort.  Who would've guessed that the music that we've all been listening to for decades can yield even more listening pleasure...I know that I was one of those "deniers". 

 

Remember what started me down this path: a CD that I couldn't listen to. and when I saw the spectrum plot, I knew what I needed to fix.  However, when I started checking out more discs that I had trouble listening to, it was kind of like the punch line of "The Sixth Sense", where the main guy figures out that something is terribly off (in the end), but we all had clues along the way, if we were willing to see (or hear) them. 

 

Since yesterday I've been going back over the tracks that I fixed - the ones that required a great deal of re-equalization so that I do the fine tuning. (I'll write more on that experience when I take a break this evening or tomorrow--I prefer to write some of this stuff down before I forget it - sort of like lessons learned.)  That process now includes rolling off extremely low frequencies below about 30 Hz for most tracks, and some I have to roll off below 40 Hz in order to balance the "thumps" o the kick drum with the overall mix.  So in reality, your comment was certainly constructive and caused me to change course.

 

So thanks for the comments--no kidding.  It's all good.

 

"May the best of your yesterdays be the worst of your tomorrows." - an old Irish saying.

[Axz Hout]

h

[DizRotus]

This thread is very interesting.

Am I comparing apples to oranges, but isn't this exactly the issue Neil Young offers as the motivation behind Pono? Also, isn't it true that high resolution digital alone will not solve the problems if the original master was done poorly? Are we naive to hope that there will be compressed mp3s for use on iPods and uncompressed dynamic recordings with quality low frequency bass information for playback on real Hi-Fi systems?

Edited March 11, 2015 by DizRotus

[Don Hills]

A few notes for anyone using Audacity for this work:

 

The later versions of Audacity (2.0.x) are better for bass analysis, because they allow a more detailed display.

 

When using the "Analyse--> Plot Spectrum" function, the window "Size:" value can be set up as high as 65536. This will provide a much more accurate display of the frequencies of the actual bass notes. The disadvantage is that you have to select a longer piece of music for analysis.

 

At the top left of the window containing the actual track, you see (part of) the track name and a downward arrow. Clicking on that area opens a drop-down menu. Selecting "Spectrogram" provides a spectrographic analysis over time for the track.

For bass analysis, go into "Edit --> Preferences --> Spectrograms" and make the following adjustments:

Set the FFT window size to 32768

Minimum frequency 0

Maximum frequency 100

Gain (dB) 10

Range (dB) 40  (experiment with setting it lower)

 

Now when you use the spectrogram, it will display frequencies from 0 to 100 Hz. You can see at a glance what appears to be actual musical content and what is just noise.

[Chris A]

Welcome Don!

 

I assume that you've been using Audacity for a while to remaster audio.  Any other observations that you'd like to share?

 

Chris

[mustang guy]

A few notes for anyone using Audacity for this work:

 

The later versions of Audacity (2.0.x) are better for bass analysis, because they allow a more detailed display.

 

When using the "Analyse--> Plot Spectrum" function, the window "Size:" value can be set up as high as 65536. This will provide a much more accurate display of the frequencies of the actual bass notes. The disadvantage is that you have to select a longer piece of music for analysis.

 

At the top left of the window containing the actual track, you see (part of) the track name and a downward arrow. Clicking on that area opens a drop-down menu. Selecting "Spectrogram" provides a spectrographic analysis over time for the track.

For bass analysis, go into "Edit --> Preferences --> Spectrograms" and make the following adjustments:

Set the FFT window size to 32768

Minimum frequency 0

Maximum frequency 100

Gain (dB) 10

Range (dB) 40  (experiment with setting it lower)

 

Now when you use the spectrogram, it will display frequencies from 0 to 100 Hz. You can see at a glance what appears to be actual musical content and what is just noise.

Hey, that's cool, and a very helpful.

 

Question. When experimenting with the range, I see it makes it more legible with lower db numbers. What is it actually doing here?

[Don Hills]

Thanks for the welcome.

 

Mustang Guy,

Take a look at the "Analyse --> Plot Spectrum" graph. Note that the highest displayed value is typically less than 0 dB.

The "Gain" setting for the spectrogram sets which value will be displayed as white (highest level) in the spectrogram display.

For example, if the highest peak on the "Plot spectrum" graph were -10 dB, setting the gain to 10 dB would mean the peak would be displayed as white on the spectrogram display. If you find that too much of your spectrogram display is white, set the gain value lower. If your peaks never get into the yellow, set the gain higher.

 

Looking again at the "Plot Spectrum" graph, look at the level of the lowest peaks of interest. The Range setting determines whether they are visible. For the example I gave, any signal that is more than 40 dB below the "gain" setting will not be displayed. So for the settings I gave, any signal in the range of -10 to -50 dB will be displayed in the colour range of white through to dark blue.

(Personally, I find the black and white display option easier to read than the colour option.)

 

For bass work I found these values gave a good display of bass that is likely to be audible, while excluding low-level rumble etc. You can experiment to find the values that best suit your usual material. (When you change one of the settings, Audacity recalculates the display so you see the effect almost immediately.)

 

You may have seen in the spectrum graphs in previous posts that the levels drop off as the frequency increases. The "Frequency gain" setting increases the gain setting as the frequency increases to make high frequencies more visible. Think of it as a visual "treble boost" comtrol. I haven't found a need to use it for bass analysis.

 

It's all in the manual:

http://manual.audacityteam.org/o/man/spectrograms_preferences.html

 

 

Chris A,

I use Audacity occasionally in my day job (telecommunications), but for audio I'm strictly a hobbyist. I use Audacity like a Swiss Army knife or Leatherman tool. There are better dedicated tools for a lot of the things I've used it for, but Audacity is free and always to hand.

One thing I did a while back was to generate a ringtone for my daugher. She liked a sound from a cartoon called "Kim Possible" (google "kimmunicator"). The ones I found online were somewhat poor - background noise etc. I recorded one off an episode (using Audacity). I then analysed the frequencies, levels and timings of the tones making up the sound and generated new tones to match.

 

My most common use is to analyse bass in the music I listen to (mostly prog and hard rock from the Beatles era to the present day). There's not a lot of older music with useful content below 40 Hz, which is as low as a 4 string bass guitar goes. Kick drum can often go lower, but the "kick" in the sound comes from the higher frequencies, 60 to 80 Hz or so. Musicians have started using 5 and 6 string basses, going down to 30 Hz, and using drop tunings, so there is some gold to be found in the lowest frequencies these days.

 

Older music production usually used an analogue high-pass filter to cut off studio rumble etc, and the LP cutting process often applied additional filtering. These filters often had gentle enough slopes that some of the low bass remained, and could be boosted per this thread. I find most modern music uses digital filters with very steep cutoff - the level "drops off a cliff" below the cutoff. Trying to boost the bass below cutoff in this case is usually futile.

[muel]

Excellent Don!  ...and welcome!

[Chris A]

Thanks Don. 

 

I've found the same things about older rock music (such as the Beatles, which contrary to my earlier comment here have responded most strongly to remastering) and also most other popular and jazz tracks--even music produced up to the present day. 

 

Perhaps I'll share more of what I've found remastering my music library--the results of which will likely upset some religiously held audiophile beliefs and paradigms about music recording formats and pedigree.  To date I've completed about 15% of my two-channel tracks--a bit more than 30% of my non-classical music library.  This includes full spectrum re-equalization of recordings--not just bass restoration. 

 

In most cases the differences are like no other remasterings that I've heard before.  The increase in detail, naturalness and realism of some of these recordings is quite startling.  Some of the resulting inverse-equalization curves built in Audacity to undo the original mastering equalization processes will likely surprise and simultaneously generate significant disgust with the realization of what we've typically been listening to.

 

Chris

[Frzninvt]

One of these (DBX 120X-DS) is all you need to remedy your lost octave of low end.  You can even fine tune the frequencies with 4 trim pots.  Always worked for me as did the Audio Control Phase Coupled Activator (Home version of the car audio Epicenter).  To restore dynamics and impact use it in conjunction with the 3BX-DS unit.

DBX 120X-DS Overview.pdf

[Ski Bum]

Can I turn up the thread rating to 11?  This thread deserves it.  Very informative and useful.  General thanks from the peanut gallery.  Carry on.