heiemo

[rachmiel]

this is an abstraakt 'solo' vocal piece based on a single phrase from a very beautiful and eerie norwegian folk song 'heiemo og nykken' about little mischievous nixies who live in the haunted norwegian forest:

www.rachmiel.org/kvr/heiemo.mp3

it's soft ... i'm getting interested in true dynamic range, rather than whispers compressed/limited to 0 dB peaks.

had some fun with reverb this time: backwards. ;-)

-----------------------------

here's another version, quite different:

http:/www.rachmiel.org/kvr/heiemo_v2.mp3

-----------------------------

enjoy both. :-)

[liqih]

you know that I do like your stuff, so allow me to be nasty now, <grin>
this heiemo is "TOO EASY", pick a beautiful a cappella voice + play a little with samples and reverbs and delays = always the same "nice" ambient stuff

[runagate]

Hmm. I hate to even say this, but audio level is entirely subjective, even that which has it's highest peak at 0 db (ditigtal).

If you don't normalize something digital, you're essential limiting the dynamic range because of the bit depth resolution of the 16-bit or whatever it happens to come out as in the mp3 you make...

Normalization is not limiting nor compressive.

That technical spiel being said, leaving something quiet for aesthetic reasons is an entirely different thing. I've downloaded this for later perusal

[rachmiel]

this heiemo is "TOO EASY", pick a beautiful a cappella voice + play a little with samples and reverbs and delays = always the same "nice" ambient stuff

everyone tells me i'm too hard and now i'm too easy? what's a boy to do!? ;-)

your point is well taken. this was actually a demonstration of a technique (for an article i'm writing) more than a piece. i'll finish the piece i was originally going to make out of this material and post it; maybe you'll like it more (maybe not). :-)

[liqih]

let me hear, <grin>

[rachmiel]

If you don't normalize something digital, you're essential limiting the dynamic range because of the bit depth resolution of the 16-bit or whatever it happens to come out as in the mp3 you make...

when you normalize an audio signal, you don't gain (or lose) any of the dynamic range or the signal to noise ratio. these both stay exactly the same. the entire signal just gets louder or softer. but i don't think that's what you're talking about ... what ARE you talking about? ;-) please elaborate.

[runagate]

If you don't normalize something digital, you're essential limiting the dynamic range because of the bit depth resolution of the 16-bit or whatever it happens to come out as in the mp3 you make...

when you normalize an audio signal, you don't gain (or lose) any of the dynamic range or the signal to noise ratio. these both stay exactly the same. the entire signal just gets louder or softer. but i don't think that's what you're talking about ... what ARE you talking about? please elaborate.

You lose dynamic range by not normalizing, because the "overhead" between your highest peak and zero is still counted towards the bit resolution being used. So if you nornalize it while it's still digital (as whatever you use almost certainly stays at 32 bit float while being edited), so before you render it to an actual .wav or .mp3 or whatever, you get that much more dynamic range resolution, which is lost if you don't. So the utlimate file rendered has less dynamic "steps" of resolution when the overhead is not utilized - the dynamic range is squished down further near the baseline of silence, so you actually hear less of the dynamics in non-normalized files.

If I could draw a picture it'd be easier to describe. That's why all recording tutorials tell you to try to record as "hot" as possible to the 0 decibel maximum, but with some wiggle room so transients don't clip, so that you're utilizing as much of the 16 bits stored as volume dynamic information as possible.

[wavephonic]

You lose dynamic range by not normalizing,

sorry just happen to be passing by, but thats news to me
there are other ways of maxing gain while keeping a good dynamic range...noise can be tripled by normalizing

put your faders to zero, record a few moments, now normalize it...have a listen, but hey I'm just a hack, maybe a revisiting of this discussion would benefit me,

most broadcast radiostations normalize...I can't stand listening to radio volume wars

btw rach...dig the track

[rachmiel]

okay, here's the version that i originally intended to create ...

heiemo_v2.mp3

this version is more of heterophony, less polyphony, like a skewed solo line.

[rachmiel]

If you don't normalize something digital, you're essential limiting the dynamic range because of the bit depth resolution of the 16-bit or whatever it happens to come out as in the mp3 you make...

when you normalize an audio signal, you don't gain (or lose) any of the dynamic range or the signal to noise ratio. these both stay exactly the same. the entire signal just gets louder or softer. but i don't think that's what you're talking about ... what ARE you talking about? ;-) please elaborate.

You lose dynamic range by not normalizing, because the "overhead" between your highest peak and zero is still counted towards the bit resolution being used. So if you nornalize it while it's still digital (as whatever you use almost certainly stays at 32 bit float while being edited), so before you render it to an actual .wav or .mp3 or whatever, you get that much more dynamic range resolution, which is lost if you don't. So the utlimate file rendered has less dynamic "steps" of resolution when the overhead is not utilized - the dynamic range is squished down further near the baseline of silence, so you actually hear less of the dynamics in non-normalized files.

understood.

but, since normalizing to 0 dB (or thereabouts) raises the volume of each sample by the exact same amount (in dB), the entire file gets louder. so, if you want a soft track, you shouldn't normalize to 0 dB, rather to -6 dB (or even less). i don't see another alternative ... though i suppose the best solution in terms of preserving bit depth would be to normalize to 0 dB and then ask (beg?) the listener to play it at a very low volume. won't happen!

http://www.kvraudio.com/forum/viewtopic ... highlight=

[runagate]

Audio normalization is the process of increasing (or decreasing) the amplitude of an audio signal. Typically normalization increases the amplitude of the audio waveform to the maximum level without introducing any distortion. If there was already distortion, it will leave it distorted and only adjust the amplitude.

Specifically, normalization applies a constant amount of gain to the selected region of the recording to bring the highest peak to a target level, usually 98% (-0.3 dB) or 100% (0 dB). This differs from dynamics compression, which applies varying levels of gain over a recording to fit the level within a minimum and maximum range. Normalization applies the same amount of gain across the selected region of the recording so that the relative dynamics (and signal to noise ratio)are preserved.

Normalization may require two passes, depending upon the software employed. A first pass would determine the highest peak, and the second pass applies the gain to the entire recording.

Normalization is often used when remastering audio tapes for CD production, in order to maximize the signal level while not changing the signal to noise ratio. It is often combined with dynamic range compression and hard limiting to increase the apparent volume of a CD. It is typically applied along with other audio and digital processing, such as dithering.

Never hurts to bone up on one's digital audio knowledge. I try to at least once a day, and I checked Wiki just to make sure I wasn't spreading disinfo, which would be worse than just not knowing.

I grabbed the new version, too, but have to go home now so I bid you happy through-composing!

[justin3am]

I like both of them but of the two I enjoyed the first one the most. I think that they are both deceptively easy, they are accessible to be sure but when you break them down they are still challenging.

As far as the "to normalize or not to normalize" question... it doesn't increase the SNR and doesn't compress peaks so it won't hurt in most applications but a low level recording with high SNR wouldn't really benefit. I would go either way with your track. You could if you wanted to but it's loud enough (with my faders at unity) that you don't have to.

[polyslax]

What's not to like? Both versions are beautiful manipulation pieces I think. I don't really care about simple/complex in terms of the finished piece or the workload to create it... the works stand on their own and sound wonderful, and there are definitely rhythmic complexities happening in my head while listening.

I think Justin nailed my understanding of normalizing... you can end up raising the noise floor depending on the source material.

[Jazzyspoon]

it's a beautiful piece, Rachmiel.