Trying to revive our dry-vs-ambient discussions

[PeterRoos]

Hi friends,

As some of you know I have been experimenting a bit in the last years with adding ambience, early reflections and reverb tails to individual instruments and groups (for my own music), as well as doing some research into the emulation of real spaces with digital (i.e. "non-real") tools.

I have always been on the "rather-too-try-than-too-wet" side of the arguments.

My latest endeavours have been again supportive for my position.

Here is a nice demo which I hope supports my "side" in the old discussion:

Classical recording, Anechoic vs Medium Large Hall, on/off

(during playback I only pressed the bypass button once in while)

Curious what you think, and if you believe that the lack of ambience in the original recording is still a problem in the mix with reverb.

Regards,

Peter
www.PeterRoos.com

[PeterRoos]

Hi Lee,

As close miced (and dry!) recordings always have more highs than when listened to at say 12-15 meters, I have rolled of the highs on the dry source first. The ratio between the (very) dry source and the reverb (indeed IRs) is not saying much, as some IRs have more energy than others (for instance in the first part of the IR, where the early reflections are). I made more demos and the setting was often in the range of dry: 0 dB and wet: -9.5 dB. But again, this can be very different per reverb type.

Recording at some distince, I agree, as you get a more even pickup of the sound, as it radiates in all directions (mostly) and thus the space act a bit as a subtle compressor.

Interesting thoughts on recording and controlling the placement using IRs and other tools (I often use TrueVerb ERs for this). I am thinking about creating an ER-only collection of IRs, with multiple positions on a single stage. I believe this can be done without going out to stages with speakers and other expensive gear.

I can't really comment on the reverb I used here, because I just want to revive the dry-wet discussion a bit, especially on the topic of early relections used for positioning.

Otherwise, I might get in trouble on this board, which I really do not want.

[Hannes_F]

As close miced (and dry!) recordings always have more highs than when listened to at say 12-15 meters, I have rolled of the highs on the dry source first.

I found the rolloff a very important indegrient of creating room together with ER and the wavetail.

However I wonder: is there any information available about how exactly a natural (physical) roll off depends on the distance? Currently I do it by ear but after all there are are three parameters ... cutoff frequency, amount and type. So a little bit more information would be appreciated.

I roughly suppose that both the cutoff frequency and the amount should be very low. This would result in an effect that basically includes the whole spectrum, but affects more the high frequencies than the lows. With other words, no sharp knee in the EQ curve since this would not be physically motivated. But this is just a guess.

Second question (maybe dumb): Isn't this rolloff already included in the IR itself? I mean, if an IR is recorded with 15 m distance, does that not already effect the frequence response?


Hannes

[wst3ae]

I found the rolloff a very important indegrient of creating room together with ER and the wavetail.

one thing to keep in mind... the energy in an enclosed space behaves in many different ways depending on many different things! This is why the Impulse Response is such a valuable tool, but since IR convolution assumes a linear transfer function, and the real world is not alway linear, there are additional steps that need to be taken. And that's before we start to add in the non-linearities associated with loudspeakers, or worse, the ear-brain interface!!!

However I wonder: is there any information available about how exactly a natural (physical) roll off depends on the distance? Currently I do it by ear but after all there are are three parameters ... cutoff frequency, amount and type. So a little bit more information would be appreciated.

There is a TON of information about how sound behaves in an enclosed space. It is divided (roughly) between large spaces and small spaces (a small space is defined as a space that is not statistically revereberent - i.e. most recording studio control rooms - ain't that ironic?)

Starting points for interesting web searches (which show my own bias for the study of small spaces, but it will provide additional pointers for large spaces) might include:
- LEDE or "Live End Dead End"
- RFZ or "Reflection Free Zone"
- Leo L. Beranek (wrote an excellent text on performance spaces)
- Harry Olson
- M. David Egan
- F. Alton Everest
- Jeff Cooper
- Philip Newell
- Chips Davis
- Diana Deutsch
- Performance Hall Acoustics
- Psycho-Acoustics
- Haas Effect

In addition, the Audio Engineering Society, Acoustical Society of America, and Syn-Aud-Con all have material, some posted for the public, some available for purchase that is very valuable.

I roughly suppose that both the cutoff frequency and the amount should be very low. <snip>

and the only answer is... you guessed it... "it depends!"

Second question (maybe dumb): Isn't this rolloff already included in the IR itself? I mean, if an IR is recorded with 15 m distance, does that not already effect the frequence response?

Absolutely! If there is a spreading of energy due to distance (it isn't really attenuation, believe it or not), then there will be less energy for the microphone to pick up, and thus the IR will account for some (most) of the apparent loss of high frequency, but there will be other factors, not the least of which is the human ear!

As an aside, back in the 1980s, when digital reverb was filtering down from the upper strata of recording studios to those of us with more modest budgets, one of the big challenges was editing the presets so that they sounded a little more "realistic"! The flip side of that was how much to return to the 2-mix. One of my mentors used to use as his rule of thumb "turn it up till it annoys you and then back it off a little." It worked!

Have fun...

Bill

Have fun...

[Hannes_F]

Bill,

thank you for your information which I will track down.

I am aware that besides any physical and acoustical facts we will always need to "trust our ears", and it will also always "depend".

Nevertheless I find this sort of information very valuable because I personally know how subjective my hearing is. My inner ear (imagination) is often even louder than my outer ear. That can be very good for creativity (playing, composing) but also very bad (all sorts of control, especially objectivity in mixing).

This subjectivity in my hearing is so strong that I can happen to ignore faults or artefacts completely until somebody points me to it. And from then on I cannot stop to hear them . Therefore the more I know the more I hear.


Hannes

[wst3ae]

thank you for your information which I will track down.

Have fun!!! As I run across other sources I'll try to send them to you...

This subjectivity in my hearing is so strong that I can happen to ignore faults or artefacts completely until somebody points me to it. And from then on I cannot stop to hear them . Therefore the more I know the more I hear.

Ah yes... isn't that an amazing phenomenon?????

I have, finally, learned to turn the analytical part of my hearing on and off - for the most part.

For a while it was very annoying because I really could not listen to music without tearing it apart! Music on the radio, the soundtrack to a movie or TV show, pretty much anything but a live performance. For some reason I was always able to go into "listener mode" in live settings.

I still fall into that trap from time to time, and sometimes when working on a project for someone else I'll slip into "listener mode" when I shouldn't, but by an large I listen appropriately.

For me, and you might want to think about this too, it came down to letting myself get lost in the music or the production when I was listening for enjoyment, and focusing on the details when I was engineering or producing or arranging. Now for me the default behavior had become analytical, so perhaps it is easier to switch from analytical to listener, I don't know.

Learning all you can about sound and music certainly enhances the analytical mode. You know to listen for specific chord changes or harmonies or noises or distortions or balances or whatever, and then you listen for them.

My current thing is listening, analytically, to TV soundtracks to see how they weave sound effects, music, and dialog together, and how the sound elements relate to the visual elements. Sometimes it takes away from the show itself, but that's ok because when I get completely lost in the show I know the writers did a wonderful job!!!

[jsp2]

I'm far from an expert in the area of effects/reverb.

Your demo was stunning! And your website has an incredible amount of equally great works.!
Rendered very believable in terms of realism.

As to your initial quarry?....My advice would be to continue with the settings you've adopted....

Whatever it is you've been doing, certainly works! ( works amazingly well!)


Awesome work!

~Jeff

[Pando]

Nice work, Peter, I think convolution shows promise.

However, I wanted to mention that in your example, convolution works with sound source that it assumes is radiating uniformly in all directions. With a static anechoic recording, the recording becomes a point source in space (or two points in case of stereo).

If you take horns for example, the bells direct the brightest spectrum of sound in one direction (not necessarily toward the microphone), and a different spectrum in another, both of which are getting reverberated and resonated in the room. If you are running a direct sound from the bell thru convolution, it doesn't have as good of an effect as if you had a horn sampled from the BACK of the instrument as well (or in all directions for that matter) which you can then run thru convolution.

This is which is why it's hard to get horns to sound good when they are sampled completely dry and ambience is added later. Things work better for instruments which are not as directional, like piano, strings, percussion, etc.

For dry vs wet in sampling, dry is definitely better for the majority of instruments, as the samples can be edited and scripting added much more easily and with realistic results, instead of fighting with the ambience burned into the samples. Loud directional instruments like horns are an exception unless someone can sample these instruments from multiple points around the instrument completely dry and then feed it all to the convolutor together with the directional information (not sure if this has even been tried before as you'd be fighting with phasing)...

Cheers

[RiffWraith]

Hi Peter - nice demo. Almost sounds like a bunch of real players...

To me, I like the reverb part(s) better, as they sound much more realistic*. THAT said, the reverb doesn't sound too realistic to my ears. It sounds like fake reverb, as opposed to a real hall. If this discussion is meant to pose the question of what's better, dry or wet, I think that's obvious. But if the question is meant to ask if samples should be recorded and released dry or wet, that still depends on alot of things....

Cheers.

*realistic here means what my ears are used to hearing. And the same would hold true for the reverse. If I was used to hearing dry recordings, then the wet versions would sound all screwed up.

[PeterRoos]

This is indeed a single artificial reverb (via convolution) added to an anechoic recording of a real orchestra. I guess it is probably not optimal to everyone's ears, but to me it supports my viewpoint that we do not necessarily need really ambient samples, but just samples with a controlled amount of early reflections (like VSL, ProjectSAM Close and a number of ProjectSAM stage samples).

This example of course lacks such early reflections, but still I find that the reverb is capable of putting the instruments further into the "hall".

I am often experimenting with ER's and still find it easier to use control really dry instruments such as Dan Dean's, than instruments with an entire hall embedded

I will make a short midi demo from my own setup with 3 versions, full dry, ER's only (TrueVerbs) and then with ER's and reverb tail.

Cheers,