Anechoic sampled instruments and reverberation

Everything and anything, but nothing about the LinuxSampler project.
User avatar
dahnielson
Moderator
Posts: 632
Joined: Wed Jan 23, 2008 11:25 pm
Location: Linköping / Tranås, Sweden
Contact:

Anechoic sampled instruments and reverberation

Post by dahnielson » Sun Mar 23, 2008 6:33 pm

I had a disturbing dream the other night that I was modeling a reverb of an anechoic room. Been occupied a little bit too much by acoustics, reverberation and instruments samples, have we?

The problem with sampled instruments and reverberation
One problem with adding reverberation to instruments samples is that said samples normally already contain reverberation to some degree. Wet sampled instruments (at least historically including many orchestra libraries) contain a high degree of reverberation, even if many of them have been edited to contain an optional release tail. The same is true for instruments sampled dry, although they are sampled in rooms with very short reverberation times (a low RT60 value) so that the recording only will contain first and early reflections whose desirable levels are figured out during instrument and microphone placement.

Adding reverb to a sample already containing reverberation will cause comb filtering to some degree depending on sound source and reverb characteristics. One solution would be to sample the instruments in an anechoic chamber (e.g. Musical Instrument Samples from the University of Iowa) or at least in a hemi-anechoic or free-field chamber (see David A. Nelson, "Acoustic Test Chambers and Environments", National Instruments, November 15, 2006).

The problem with acoustic instruments and radiation
But anechoic recording has its own set of limitations. For instance microphone placement in relation to the radiation of sound by acoustic instruments (Scott Wilkinson, "Resonance and Radiation", Electronic Musician, September, 1994) need to be taken into account. In an anechoic chamber sound radiated as non-direct sound will not be reflected back and recorded by the microphone like in a studio or auditorium. This is especially troublesome with bowed strings having shifting radiation patterns depending on the tone being played.

Image
Principal sound radiation patterns for the cello.

Conclusion
An omni-directional source, like a starter pistol or carbon arc, is regularly used as impulse during the recording of a room impulse responses. After convolution with a sample the output will be analog to the sample being played back as an omi-directional sound source in the room. To be precise: the impulse response will contain all characteristics of the impulse source, room, microphone and signal chain. An instrument that naturally focus its sound forward (as principal radiation pattern), and being mic'ed accordingly when recorded, will then radiate the same principal sound level and frequencies in all directions in the reverberation model. This is however countered under normal circumstances by subjectively balance the dry signal, representing the direct sound, with the wet signal produced by the reverb in the mix.

With a synthesized impulse response it would be possible to take the orientation of the sound source and radiation pattern into account (Perry R. Cook and Dan Trueman, A Database of Measured Musical Instrument Body Radiation Impulse Responses, and Computer Applications for Exploring and Utilizing the Measured Filter Functions, Princeton University, 1998) and filter the dry input to simulate the non-direct sound being radiated and reflected back to the microphone.

(Permanent URL.)
Last edited by dahnielson on Sun Mar 23, 2008 7:43 pm, edited 1 time in total.
Anders Dahnielson

Ardour2, Qtractor, Linuxsampler, M-AUDIO Delta 1010, Axiom 61, Korg D12, AKAI S2000, E-MU Proteus 2k, Roland R-5, Roland HP 1300e, Zoom RFX-1000, 4GB RAM x86_64 Intel Pentium Dual 1.80GHz Gentoo Linux

User avatar
Consul
Moderator
Posts: 189
Joined: Wed Jan 23, 2008 11:19 pm
Location: Port Huron, Michigan, USA
Contact:

Re: Anechoic sampled instrument and reverberation

Post by Consul » Sun Mar 23, 2008 7:14 pm

http://bram.smartelectronix.com/plugins.php?id=5 :mrgreen: :mrgreen: :mrgreen:

The Larry Seyer Drum Library for Gigastudio was designed using samples of completely dry drums which are then passed through impulse responses of rooms, transformers, channel strips, and any number of other things (he even has an IR of a trampoline). This is the approach I plan to take when I can get around to creating my own drum library for open source release.
Darren Landrum

User avatar
dahnielson
Moderator
Posts: 632
Joined: Wed Jan 23, 2008 11:25 pm
Location: Linköping / Tranås, Sweden
Contact:

Re: Anechoic sampled instrument and reverberation

Post by dahnielson » Sun Mar 23, 2008 7:31 pm

ROFLMAO.

BTW, the worlds largest anechoic chamber is located at Edwards Air Force Base, the Benefield Anechoic Facility as apart of the Air Force Flight Test Center. It's however not an acoustic anechoic chamber but intended for radio frequency testing. While Orfield Laboratories anechoic chamber is "The quietest place on earth" measured at (minus) -9.4 dBA.

For some insight into the construction of one, take a look at this article about McIntosh Laboratory's acoustic anechoic chamber built in 1979 when beards still was legal!
Anders Dahnielson

Ardour2, Qtractor, Linuxsampler, M-AUDIO Delta 1010, Axiom 61, Korg D12, AKAI S2000, E-MU Proteus 2k, Roland R-5, Roland HP 1300e, Zoom RFX-1000, 4GB RAM x86_64 Intel Pentium Dual 1.80GHz Gentoo Linux

User avatar
Consul
Moderator
Posts: 189
Joined: Wed Jan 23, 2008 11:19 pm
Location: Port Huron, Michigan, USA
Contact:

Re: Anechoic sampled instruments and reverberation

Post by Consul » Sun Mar 23, 2008 8:31 pm

I wish I had access to an anechoic chamber for my sampling needs. Maybe once I transfer to university, I'll have one there. That won't be for over a year, though. I might have to make do with some creative improvisation.
Darren Landrum

User avatar
dahnielson
Moderator
Posts: 632
Joined: Wed Jan 23, 2008 11:25 pm
Location: Linköping / Tranås, Sweden
Contact:

Re: Anechoic sampled instruments and reverberation

Post by dahnielson » Sun Mar 23, 2008 9:01 pm

A full anechoic chamber (room sized that is) is expensive to build because of the floor suspension (take a look at the McIntosh article and see what I mean) and probably be overkill for sampling purposes. But a hemi-anechoic chamber is doable and even a free-field chamber (which is more or less equivalent to a highly treated music studio) would be purposeful. The trouble, as always in room design, would be the lower end (the room mode not being affected by anechoic wedges) why the room size in addition to the wedge material dictate the limit of anechoic sampling in the bass end.

The handy chart of orchestra instruments tell us we need at least 5.22 m = 344 m / (2 x 33 Hz) to the ceiling, if we will be dealing with the double bass, giving us a room with the following properties:

Ratio: 1:1.4:1.9
Area: 72.480744 m2
Volume: 378.34948368 m3
Dimensions: 5.22 x 7.308 x 9.918 m
Freq.: 32.9501915709 : 23.5358511221 : 17.3422060899 Hz

Image

Eckel's standard wedge is based on the original geometry established by Beranek and Sleeper at Harvard during WWII.

Image
The original wedge.
Data on the performance of five different types of structures for use in echo-free (anechoic) chambers are presented. The best one of these structures is shaped like a wedge and manufactured from glass fibers held together by a binding agent. When mounted in the room, the wedges are spaced out several inches from the walls, and the dihedrals of adjacent units are turned through 90°.

Generalized specifications for the optimal design of structures made from glass fiber wedges are presented in terms of either (a) lowest frequency at which 99 percent or better absorption is desired or (b) maximum depth of treatment which may be installed in the room. The application of these specifications to two rectangular rooms is shown and inverse square law measurements performed in the two completed chambers are presented. In the larger chamber the deviations are within ±0.3 db out to 10 feet and ±1.0 db out to 30 feet from a point source of sound. In the smaller, the deviations are within ±1.0 db out to 10 feet.
Leo Beranek and Harvey P. Sleeper, Jr., "The Design and Construction of Anechoic Sound Chambers", Journal of the Acoustical Society of America 18 (1946): 140-150.
An anechoic wedge is designed to provide a normal-incidence sound absorption coefficient greater than 0.99 for all frequencies down to its design cutoff frequency. Reflected sound from such a wedge is attenuated 20 dB or more. An anechoic wedge is typically one-quarter wavelength long at its cutoff frequency. For example, a 100 Hz cutoff wedge is usually approximately 36 inches long. The design and production of effective anechoic wedges is an extremely difficult task and is best left to acoustic chamber manufacturers.
David A. Nelson, "Acoustic Test Chambers and Environments", National Instruments, November 15, 2006.

For more information see Brüel Acoustics, Anechoic Chambers, Technical Review 96-02. Especially the flow resistor and the Cremer units, a substitute improvement over the Beranek-Sleeper wedge. (What Nelson calls "free-field chamber" is called "open-field chamber" in Brüel, "free-field chamber" is a full-anechoic chamber in Brüel.)

One of the most important documents regarding general studio design is without doubt BBC Research, Guide to Acoustic Practice, 2nd edition, 1990.

Also Ethan Winer's "Recording Spaces", EQ Magazine, June 2004, has a good explanation of comb filtering, discussion of hard floors (which is related to hemi-anechoic and free-field rooms) and put the radiation of acoustic instruments into context.

The question is: How do we construct a good chamber with free-field condition over a reflecting plane on limited budget and with limited resources? Ideally one that anyone with some vested interest can build to use for sampling.
Last edited by dahnielson on Tue Mar 25, 2008 1:58 pm, edited 17 times in total.
Anders Dahnielson

Ardour2, Qtractor, Linuxsampler, M-AUDIO Delta 1010, Axiom 61, Korg D12, AKAI S2000, E-MU Proteus 2k, Roland R-5, Roland HP 1300e, Zoom RFX-1000, 4GB RAM x86_64 Intel Pentium Dual 1.80GHz Gentoo Linux

User avatar
Consul
Moderator
Posts: 189
Joined: Wed Jan 23, 2008 11:19 pm
Location: Port Huron, Michigan, USA
Contact:

Re: Anechoic sampled instruments and reverberation

Post by Consul » Sun Mar 23, 2008 10:16 pm

Well, Larry didn't have an anechoic chamber handy when he recorded his library. It was just him, Pat Mastelotto, Pat's drums, and some very close-miking, one drum at a time (they didn't set up any kits).

Another interesting trick is exemplified in the portable vocal booth:

http://www.digitalprosound.com/articles ... p?id=89503
http://www.hometracked.com/2006/12/17/p ... cal-booth/

A combination of this and some old bales of insulation sitting in the corners of my basement will likely be the solution to my problem.
Darren Landrum

User avatar
dahnielson
Moderator
Posts: 632
Joined: Wed Jan 23, 2008 11:25 pm
Location: Linköping / Tranås, Sweden
Contact:

Re: Anechoic sampled instruments and reverberation

Post by dahnielson » Sun Mar 23, 2008 10:23 pm

Yes, a couple of SuperChunks in the corners and some cloth covered 703 slabs (or similar) will approximate a free-field chamber*. A good idea would also be a couple of clouds or hangers in the ceiling (if the ceiling height allow it).

* In case anyone wants to know: according to National Instruments glossary a free-field chamber is "a special test chamber with a hard, reflecting floor whose other surfaces are highly sound-absorptive but not anechoic, thereby approximating free-field conditions" and according to ANSI acoustical terminology article 6.06 a free-field is a "field in a homogeneous, isotropic medium free from boundaries" which NI's glossary clarifies "that is, the wavefronts spread out continuously in a manner analogous to ripples on a pond without reflections." :ugeek:
Last edited by dahnielson on Tue Mar 25, 2008 1:55 pm, edited 4 times in total.
Anders Dahnielson

Ardour2, Qtractor, Linuxsampler, M-AUDIO Delta 1010, Axiom 61, Korg D12, AKAI S2000, E-MU Proteus 2k, Roland R-5, Roland HP 1300e, Zoom RFX-1000, 4GB RAM x86_64 Intel Pentium Dual 1.80GHz Gentoo Linux

User avatar
dahnielson
Moderator
Posts: 632
Joined: Wed Jan 23, 2008 11:25 pm
Location: Linköping / Tranås, Sweden
Contact:

Re: Anechoic sampled instruments and reverberation

Post by dahnielson » Sun Mar 23, 2008 10:48 pm

Consul wrote:Another interesting trick is exemplified in the portable vocal booth:

http://www.digitalprosound.com/articles ... p?id=89503
http://www.hometracked.com/2006/12/17/p ... cal-booth/

A combination of this and some old bales of insulation sitting in the corners of my basement will likely be the solution to my problem.
Cool. The idea is similar to commercial products like RealTraps Portable Vocal Booth, The MIC THING, sE Electronics Reflexion Filter and Auralex Xpanders.
Last edited by dahnielson on Mon Mar 24, 2008 3:52 pm, edited 3 times in total.
Anders Dahnielson

Ardour2, Qtractor, Linuxsampler, M-AUDIO Delta 1010, Axiom 61, Korg D12, AKAI S2000, E-MU Proteus 2k, Roland R-5, Roland HP 1300e, Zoom RFX-1000, 4GB RAM x86_64 Intel Pentium Dual 1.80GHz Gentoo Linux

User avatar
Consul
Moderator
Posts: 189
Joined: Wed Jan 23, 2008 11:19 pm
Location: Port Huron, Michigan, USA
Contact:

Re: Anechoic sampled instruments and reverberation

Post by Consul » Mon Mar 24, 2008 1:11 am

The DIY solutions cost about a tenth of the commercial solutions. They cost even less if you, like me, already have a mess of rockwool and fabric hanging about. :D

I have an idea for the "Stand of a Thousand Things" to sit in the middle of my room. At chair height, a mess of finger- and hand-activated piezo pads for playing drum parts, and at standing height, a Cube O' Rockwool for vocals. Somewhere in there will be another Cube O' Rockwool with a mic to accommodate didgeridoo recording. I'll probably also build on a bunch of hooks to hang cables on, and probably a couple of guitar hooks as well.

This will be after finishing up the treatment of the rest of the room, using some of the aforementioned mess of rockwool I have hanging about.
Darren Landrum

User avatar
dahnielson
Moderator
Posts: 632
Joined: Wed Jan 23, 2008 11:25 pm
Location: Linköping / Tranås, Sweden
Contact:

Re: Anechoic sampled instruments and reverberation

Post by dahnielson » Mon Mar 24, 2008 1:16 am

Yes, when it comes to acoustics I wouldn't spend any money on ready made solutions (of which half is usually scams like "acoustic foam" products costing ten times more than a piece of 100 kg/m3 rockwool) but build my own too. Just pointed out some similar commercial products based on the same idea.
Anders Dahnielson

Ardour2, Qtractor, Linuxsampler, M-AUDIO Delta 1010, Axiom 61, Korg D12, AKAI S2000, E-MU Proteus 2k, Roland R-5, Roland HP 1300e, Zoom RFX-1000, 4GB RAM x86_64 Intel Pentium Dual 1.80GHz Gentoo Linux

Post Reply