Getting consistent audio song level across registrations

[pjd]

Hi --

Jim's post about SSS  -- and John's reply -- prompted me to ask about something which is a long-standing issue with me.

I play over WAV/MP3 backing tracks, AKA "Audio Songs". I use one registration per song in which the registration recalls the Audio Song and the voice(s) to use when jamming over the Audio Song. The registration saves the Audio Song level, RIGHT1 level, etc. I try to balance the Audio Song level versus, the RIGHT1 level, etc. so as to play over the Audio Song and have the solo speak at an appropriate relative volume.

My issue is obtaining a consistent Audio Song level across all of my songs/registrations. By this, I mean each WAV/MP3 plays at the same consistent final volume, i.e., the same signal level to the mixer/speakers.

To this point, I've set "the song volume" by ear and get reasonably consistent results. However, I fear that ear fatigue eventually kicked in and I still have a few occasions where the next audio song is too soft or too loud.

I wish there was a VU or peaking meter to give a visual read-out of Audio Song level. The meter in the Master Compressor doesn't help because the Audio Song signal bypasses the Master Compressor and EQ. I've done a few experiments using an external mixer with its own built-in meter, but I'm not quite satisfied yet.

Sooooooooo, any suggestions? Thanks! Always lots of good ideas here.

-- pj

[Fred Smith]

I agree. I've always done it by ear, with less than ideal results.

The only solution I've found is to output to a recording device that has a meter.

I'm interested in any other solutions.

Fred

[DrakeM]

I have a program .. Cool Edit Pro .. which is similar to Audacity, so maybe you can get it and use it. Cool Edit Pro is no longer available.

There is a feature under Amplitude called Dynamics Processing.
This opens a menu up with several preset settings. I use one called "limit hard -6db". I apply this to all my MP3 recordings and then I use a Normalize feature ( which Audacity also has) to the track.

The "limit hard -6db" feature does have a visual graphic which shows it is compressing the track's sound. I don't know if I am using the correct setting but the tracks sound fine to me after I apply it.

So, have you checked out Audacity yet?

Regards
Drake

[JohnS (Ugawoga)]

Hi
Another useful tool for using consistancy I have found Is MASTERMATCH from IK Multimedia.

You can put up to three songs from the Genre you want to get near to like chart hits etc or any top song that you fancy.
You then press a button to take a reading of the three songs and then play your song In a sequencer such as Cubase.
After playing a section or all of your song the program also reads that .
After all of that press Mastermatch and all the sliders have a little move about on their own and bingo ,you have a song that is of the quality of the studio one.
It all depends on the quality of your song though.
Keyboards such as the Genos have digital sounds and you cannot go far wrong
I am finding Mastermatch quite useful.
You could make a template in Mastermatch and you could match all wave files to It.
I do not know but you may have a free  trial on Mastermatch for so long.

It is a difficult thing to get consistancy as  some songs are very loud and some soft at the same decibel level.
It Is all in that magic stuff called experience and I am still learning.

All the best
John

[pjd]

Hi --

Thanks for the suggestions, folks!

Good point about dynamics processing, Drake. I have Audacity and various plug-ins. I suspect that I still need to set the Audio Song level appropriately on the Genos. I'll definitely continue with the meter approach, Fred. MasterMatch sounds interesting, John. I'll check it out.

Have a good weekend, everyone -- pj

[JohnS (Ugawoga)]

Hi
I do notice that if you have sound levels matching it cuts a lot of miss matching out.

For me some of the the fills glitch if you change from one registration to another.
If you hit a fill with a pedal asigned and then change you get a glitch with the fill.
It is like a blip in the works.
Otherwise you have to wait until the fill has finished and then change at the speed of light back to putting your fingers on the keyboard with a different sound
The only way i see around this is to create a fill on a pad.
It does not happen on all styles though.

All the best
john

[8t8KEEZ]

One thing I've noticed -

Let's say R1 Brass on Registration 2 / Volume set to 60, because I don't want the brass blaring ....

Let's say R1 Piano on Registration 1 / Volume set to 100, because it just needs that extra little punch ...

So - I'm playing Reg 1, piano, and I have to transition to Reg 2, to play the brass ... and at the end of my phrase, I'm sustaining with the piano .... and I punch Reg 2... The brass comes blaring in... UNTIL I release and continue to play... OR vice versa ...

What I've found I am needing to do is to make sure the volumes of everything from registration to registration is fairly consistent, otherwise, I can expect strange results when moving from one registration to another...

In your situation - unless you "normalize", you'll always get different results ... No ONE audio MP3 is recorded the same... In my opinion, your "ear" is probably the best tool... BUT, you take it a step further, and balance all of the instruments across ALL of the registrations you use for that MP3 ... AND of course, this puts you in the neighborhood ... and from there, use "live control" to manually adjust on the fly...

I LIKE how you are doing this though ... it's a great idea for what I want to do.

[pjd]

Hi --

Just a few extra details...

I use one registration per song. So, if I change registrations, I change songs. Not everyone use this approach and, yeah, I have experimented with multiple registrations per song. 

Anyway, in addition to the WAV assigned to the registration, I also assign a customized style where the OTS buttons select alternative voices for  LEFT, RIGHT1, ...  The style is silenced by setting the STYLE part to zero in the mixer. Being somewhat OCD, I also set the style zone to the lowest key. No point in triggering the style/accompaniment if I'm not going to use it!

BTW, I do get most of what I want by ear, but need an objective cross-check (level metering). An external mixer (with meter) does the job although I'm looking into other alternatives, i.e., a software app or whatever. The mixer is easier to set up consistently -- all the danged parameters (knobs) are immediately visible.

Thanks for all of the thoughts, folks -- pj

[XeeniX]

Hi PJ,

Instead of adjusting all the audio song levels  in your registration would it be an idea/solution to do an automated audio level adjustment?


I have a huge music collection of carefully sorted and tagged mp3 tracks from quite a lot of different albums. This resulted in me at first go to the volume knob to adjust volume louder or softer every few songs since the recording volume levels differ per album usually. My solution was to use an external program where I could set the db level for all mp3's to the same level (I used 95db). It can do this per song or per folder so batch adjusting and do all of them in one go is not a problem at all. To prevent clipping of sound (some mp3's had different peak levels) you can use the option (and I always do) <don't clip when doing track gain>. The only downside is that it only works with mp3 and not with teh wav format as far as I know. You could however convert the wav's in your backing track collection to mp3 if you like. After replacing the previous audio files with the mp3gained ones your registration would still work unless you had to convert wav to mp3 in which case you could use Murray's Registration manager to do a batch rename from .wav to .mp3  You can even do a non destructive track analyses of the different db levels beforehand than use track gain to set them all to the same db level. The don't clip can be found in the options menu btw.

The program is called mp3gain, is open source (free) and can be downloaded here:
http://mp3gain.sourceforge.net/


It has worked wonders for me so I hope it helps


regards,
Peter

[attachment deleted by admin]

[Kaarlo von Freymann]

Hi --

.....To this point, I've set "the song volume" by ear and get reasonably consistent results. However, I fear that ear fatigue eventually kicked in and I still have a few occasions where the next audio song is too soft or too loud.
......Sooooooooo, any suggestions? Thanks! Always lots of good ideas here.

-- pj

Hi pjd,
ear fatigue is a problem. I use a precision dB meter when programming registrations  in my studio to get the volume set correctly.  The meter must of course always be in the same location  (same distance from the speakers) and it helps if it also has a peak hold function. In a performance situation the meter  helps setting your volume compared to the noise your audience makes. Members of the audience of course have their own preferences and hearing deficiencies. Most every time someone comes and says "I cannot hear what you play, turn up the volume"  which if you have no meter will easily make you increase the level too much. With the meter you can always say, sorry I am already as loud as I am allowed to be.

Cheers
Kaarlo

[pjd]

I use a precision dB meter when programming registrations  in my studio to get the volume set correctly. 

Thanks, Kaarlo. I was wondering if an external dB meter could be used in this way and, indeed, you've done it!

All the best -- pj

[Marcus]

When I use Audio Multi Recorder on the Genos, I always use "normalize" within the sub tracks and the final multi track render. Consistent audio at the highest audio quality levels.

From within the Genos "Audio Multi Recorder" interface, one can access "Menu" and then, "Import". From "Import" you can now access any wav or mp3 audio file from within the Genu User or USB drives. After the wav or MP3 is imported, "Normalize"(bottom right) is highlighted and available. Choose, normalize "Main Track", and the audio file with be normalized at a consistent audio level. You can over-write the original or rename the "normalized" version.

Consistent audio levels is a must, just as I try to keep with the 100/127 volume level standard within the Genos mixing console for all the tracks and parts. All my custom styles, voices, Multipads and registrations saves maintain consistency with volume levels that work well with each other.

When I used to make lots of my own expansion custom voices and drum kits, each source sample was "normalized" for consistency and then tweaked within the element(s) for the final sound set. In the same way "normalizing" all your audio sources will at least get you in the ballpark for a consistent sound level and then tweaked to suit. Yes, tweak over a few days, because ear fatigue is a factor, however your own ears is the best tool to what sounds best in the mix or final song.

Marcus

Marcus

[pjd]

Hi folks --

I'm making progress. Some quick info.

Normalization is like dialing up the volume knob until the highest peak hits a target level. Everything else in the track is dialed up (amplified) by the same gain factor. That eliminates clipping.

However, our ears perceive loudness differently. The recommended contemporary approach measures loudness cumulatively across an audio region or track. Then, a mastering tool dynamically applies gain across the track to achieve the target loudness. John mentioned MasterMatch and Peter mentioned mp3gain. Unfortunately, mp3gain doesn't work with WAV because it uses MP3 metadata to non-destructively adjust the loudness. WAV doesn't have the equivalent metadata.

I'm going to experiment with iZotope Ozone 8 Elements, which applies processing ("the Maximizer") similar to MasterMatch. I've met some of the iZotope folks -- they're a local company. Plus, I got a sweet deal on a Magix Audio Studio + Ozone Elements bundle. I needed to update Audio Studio and essentially got Ozone Elements for free. (Check the Magix site for occasional deals.)

I kind of proved this to my ears by playing 10+ tracks through a meter. They all peaked at roughly the same level, but I perceived a few tracks as louder than others. Example: a track with punchy drums (Mustang Sally) versus a horn-heavy tune (Knock On Wood). The meter transiently hits the target peak with Mustang Sally, but consistently hits the target peak and stays there during the horn parts in Knock On Wood. I perceive Knock On Wood as louder although both tracks are normalized.

Hope this helps! Thanks for all of the comments -- pj