GLL files, and what EASE Focus does with them

AFMG replaced a table with an object model, and that's most of the story. The old EASE format described every loudspeaker, however many drivers and amplifiers it had, as one point source with one balloon. A GLL starts from a box type: a cabinet with its geometry, weight and centre of gravity, a set of acoustic sources placed and aimed inside it, and the inputs and filter groups that feed them. The bigger objects are built out of box types.

• Maximum output is a voltage. Maximum SPL comes from the measured sensitivity, the spectrum of the signal you pick at calculation time, and the source's Program V_rms (1 minute) rating. The Peak (1 s) and Continuous (100 h) voltages are stored too, and SpeakerLab labels them information only. So the manufacturer's choice of that one voltage sets every max-SPL figure you'll see, which matters later.
• Phase is kept, and the time of flight is taken out of it. The propagation delay to each measurement point is stored separately, which flattens the phase slope and lets 1/24-octave spacing describe it further up the band than it otherwise could. Phase is what makes two boxes interfere properly in a prediction, and it's the thing the old format didn't have per driver.
• What you can change is up to the author. The project decides which parameters the end user sees: splay angles, box counts and mixes, filter settings, input configurations. A line array in a GLL can only be reconfigured as far as its manufacturer allowed, which is why two GLLs for similar boxes can behave very differently in Focus.

The old SPK format kept a 5° balloon in third-octave bands from 100 Hz to 10 kHz (older EASE versions used 10° or 15°). A high-resolution GLL source keeps 1/24-octave points over the whole audio band. Counting magnitude and phase at every direction on the sphere, that's about 11 times the data per source at the same 5°, and a multi-way box has a set of those per driver.

Two rules about phase, side by side

The SpeakerLab guide gives two criteria, both from the same idea: between neighbouring data points the phase mustn't turn by more than a quarter cycle. Both depend on r, the distance between the rotation point and the source's acoustic centre, because that offset is what makes phase spin as the box turns.

frequency: r ≤ c ÷ (4 · Δf)    angle: Δθ ≤ asin(c ÷ (4 · f · r))

• Frequency. At 8 kHz neighbouring 1/24-octave points are 231 Hz apart, so r must stay under 0.37 m. That's the guide's own worked example, and the build checks it.
• Angle. Put 5° into the second rule at 8 kHz and r must stay under 12 cm. The white paper elsewhere accepts a source 0.25 m from the rotation point at a 6 m distance, which answers a different question (the path to the microphone); at that offset the angular rule wants 2.5° steps, not 5.
• What the two say together. A 5° GLL is good to 8 kHz when each driver was measured about a point close to its own acoustic centre, which is why the white paper asks for that for anything filtered separately. Above a few kHz, a GLL's phase is as good as the lab's choice of rotation point, and nothing in the file tells you which choice was made.

The same file doesn't work everywhere, which AFMG's own FAQ says up front. The differences are in what kind of object each program accepts and whether it insists on a paid-for authorisation.

Manufacturers version their own files, each in their own way. RCF's 2025 release names files like GTX10_25_v1_F3, with Focus 3 compatibility marked per model. Q-SYS number their library 4v01 and 4v02 with a separate revision per product family. Proel once released "EASE Focus GLL 3.0", which was for EASE Focus 2. A version number on a download is the manufacturer's, and it tells you nothing about which Focus it's for unless they say so.

• Look in Focus. File → Manage System Definitions (Ctrl+Alt+M) lists what's installed, and its search covers manufacturer, system and GLL version. Hover the blue i for the definition's unique ID and any plug-ins it carries. Compare that with the manufacturer's download page before a show, not during it.
• Import copies the file. Focus keeps its own copy of every GLL you import, so the download can go afterwards. It also means updating a box is a fresh import, and the old one stays in the database until you remove it.
• Projects bring their own GLLs. Saving an .fc3 packs the GLL and DLL files it uses into the project, and opening someone else's project lists the system definitions you don't have and offers to save them into your database. Read what they are before pressing Save All: that's how an older GLL arrives on your machine attached to a project. I haven't tested which copy wins when two versions of one box are installed.
• The GLL Viewer shows a file's information, authorisation and exposed settings without installing it into anything, and is the quickest way to look inside a GLL somebody emailed you. It doesn't open GLL + DLL packages.

• Max SPL isn't comparable between brands. It's computed from whichever voltage the manufacturer entered as Program V_rms, and they don't agree on what that should mean. RCF's technical note tells you to take 12 dB off in AFMG software for a realistic continuous level. Q-SYS's 4v01 library changed its max SPL to continuous RMS (AES2 pink noise, 1 m, after a minute, with protection running). Focus then adds its own choice of input signal: Program (IEC 268), Music (EIA-426-B) or pink noise (AES2), each with a different spectrum. Read each manufacturer's note before putting two brands' max SPL side by side.
• The preset you'll run may not be in the file. Line array presets come from the GLL; if it has none, Focus offers its own J-array, lined and curved shapes, which are geometry and not the manufacturer's processing. Filter settings appear only when the GLL exposes input filters. A ground-stacked sub left on its flown input configuration is 6 dB light in Q-SYS's case, and Q-SYS also warn that gain changed in EASE's object properties doesn't reach the individual boxes (the filter view does).
• Limits are the manufacturer's model of the rigging. Weight and angle limits, minimum box counts and warnings are whatever the author wrote in, and Focus shows them in the status bar. Accessories outside the model aren't counted: Q-SYS note their pull-back hardware isn't in the simulation. A green status bar is the GLL agreeing with itself, and a rigging calculation is still somebody else's job.
• Focus models direct sound only. No reflections, no shadowing, no wind; air absorption to ISO 9613 is included. Elements within one array are always summed coherently, which can overstate interference on tall arrays outdoors. Between separate arrays and loudspeakers it's coherent below a set frequency and power summation above, 250 Hz by default, crossfaded over an octave (177–354 Hz at the default; the guide's 1 kHz example runs 707 Hz to 1.41 kHz). Two hangs close together will interfere in the room more than the default shows them doing.
• The venue gets tidied. Focus has no geometry import: you type coordinates or trace over a picture. Its audience zones are plan rectangles with a height profile, and AFMG document a 2 m minimum and a 45° maximum slope. The 2 m applies to width and is enforced silently: when I wrote 57 audience planes from a real theatre into a project for ArrayCAD, 45 of them came back from EASE Focus 3.1.260 widened to 2 m, centres unmoved, nothing on screen.

• SPK (1990). EASE's SpeakerBase: one point source per box, 5° balloon, 21 third-octave bands from 100 Hz to 10 kHz. Small, easy to produce, and unable to describe a multi-way box, a DSP preset or a mechanical configuration. SpeakerLab still imports it, and the ASCII .xhn.
• DLL (EASE 4.0 on). A programmed Windows library per product, the most flexible option and tied to the one program it was written for. Before GLL a line array reached EASE as one of these, configured in the manufacturer's software or in EASE Focus. GLL object definitions grew out of them.
• CLF. The Common Loudspeaker Format, an open alternative used by programs outside AFMG's: fixed tables, a 10° grid in .cf1 and 5° in .cf2, octave and third-octave bands, and phase only as third-octave averages per AFMG's comparison. AFMG's June 2014 count put GLL at 122 manufacturers and CLF at 26, which is a rival's census of a rival and worth weighing as such. A free CLF viewer is published by the CLF group.
• Going down is allowed. SpeakerLab Pro exports a GLL's sources to the simpler formats (EASE .xhn, CADP2, Ulysses UNF and CLF 1 and 2). Nothing goes the other way: a GLL made from an SPK is a point source in a nicer file.
• Manufacturer-only tools such as d&b ArrayCalc, L-Acoustics Soundvision and Meyer MAPP 3D run on their makers' own data and don't load GLLs. A manufacturer with its own tool may or may not publish GLLs as well, and when it does, check whether they're authorised for Focus or only for EASE. ArrayCAD moves venue geometry between ArrayCalc, Soundvision and an EASE Focus .fc3; the loudspeakers stay with each program.

For anyone measuring boxes rather than predicting with them: pick a grid and a symmetry, say how far the source's acoustic centre sits from the rotation point and how high you need the data to hold, and it counts the files SpeakerLab will expect and checks both phase rules.

• AFMG, GLL (Generic Loudspeaker Library): A New Standard for Measuring and Storing Loudspeaker Performance Data, engineering white paper, rev. October 2007. The object hierarchy, the two-step build, the measurement guidance (far field, rotation point, neighbours in place, 5° and about 1/24 octave), the DLL lineage and what the model leaves out.
• AFMG, EASE SpeakerLab User's Guide, rev. 1.0, December 2016: §1.6 on SPK, DLL and GLL and the .xgll project, §1.7 on authorisation, §2.7.1–2.7.2 on resolution, file names and symmetry, §2.7.4.4 on maximum voltage, §2.9.2 on the .author file, Appendix B on .xglc, and §8.5.4–8.5.5 for the two phase criteria and the 230 Hz / 0.37 m example checked above.
• AFMG, EASE Focus 3 User's Guide, rev. 3, February 2019: system definitions, import and management (§2.7–2.9), presets and limits, the calculation model and the summation crossfade (§4.3). It predates 3.1.310, so the .xglc and installer changes come from the EASE Focus upgrade history on afmg.eu, read to 3.1.321 (28 October 2026).
• AFMG, FAQ entries "Can I use the same GLL loudspeaker file in all AFMG modeling programs?" and the customised-installer answer for manufacturers; the GLL Loudspeaker File Format page for the comparison with CLF and UNF and the June 2014 database counts; EASE Focus 3 License Fees, v5.0, 2 February 2023; the EASE GLL Viewer download listing.
• Manufacturer notes: Q-SYS PL Series EASE Guide (library 4v01/4v02) for max SPL, input configurations, gain and accessories; RCF's technical note of 22 April 2025 on its version _25 GLLs and the 12 dB correction; Proel's "EASE Focus GLL 3.0" release for Focus 2 as reported by ProSoundWeb.
• Not read directly: Feistel and Ahnert, "Modeling of Loudspeaker Systems Using High-Resolution Data", JAES 55 (7/8), 2007, and Feistel, Ahnert and Bock, "New Data Format to Describe Complex Sound Sources", AES 119th Convention paper 6631, 2005, both cited through the white paper; AES56-2008, cited through the SpeakerLab guide; the CLF grid sizes, from ODEON's documentation as summarised in search results.
• My own: the .fc3 container and the silent 2 m widening, from reverse-engineering the project format against EASE Focus 3.1.260 for ArrayCAD on 3 August 2026.

Assembled 3 November 2026 with AI assistance. The figures AFMG publish that can be recomputed (the 1/24-octave spacing and 0.37 m limit, the summation crossfade, the SPK band count, the per-model fee) are checked at build time, and the 12 cm grid finding is computed from AFMG's own criterion. Those checks catch arithmetic, not a wrong source. I haven't taken a compiled GLL apart and AFMG don't document the format, so everything about its insides comes from AFMG's own description. The AFMG guides are 2016 and 2019 editions; current versions may word things differently.

Companion pages: inverse square law and air absorption for the ISO 9613 losses Focus adds, polarity, phase and time alignment for coherent against power summation and why phase data matters, and loudspeaker impedance for the rated impedance and 2.83 V sensitivity a GLL's figures are referred to.