Glass Wool vs Rock Wool for Soundproofing: Which Absorbs Better?

Glass wool and rock wool are both mineral-fiber insulants, both inorganic and non-combustible, and both used across building and industrial acoustics. But when a specifier asks "which one is better for soundproofing," the honest answer is that neither material, on its own, is a sound wall. Both are primarily sound absorbers inside a cavity, not sound barriers across an assembly. This guide separates the two jobs, compares glass wool and rock wool on the metrics that actually matter — absorption coefficient, NRC, density, temperature and fire — and gives a scenario-based selection matrix for walls, studios, ceilings, HVAC and plant rooms.

(See our glass wool and rock wool guides, and the broader glass wool vs rock wool: 7 differences comparison.)

1. The Core Misconception: Absorption Is Not Transmission

For an acoustic consultant, room designer or noise-control contractor, the first decision is not "does rock wool beat glass wool," but to split the problem into two different physical jobs:

• Absorption (porous): reducing standing waves, reverberation and panel radiation inside a cavity.
• Transmission (barrier): stopping energy from a source room crossing the element into a receiving room.

Glass wool and rock wool belong almost entirely to the first job. Their NRC and alpha(f) describe how much sound energy the material or assembly dissipates, not how many decibels a single layer blocks.

A real wall's sound isolation comes from a correctly sequenced system: continuous air sealing, sufficient surface mass, mass de-coupling across the two leaves, resilient or staggered studs, ceiling and floor, a deep but not over-stiff cavity, and only then filling the cavity. Published steel-stud assembly data make the point clearly: with the same glass-fiber fill, a single-board thin-stud wall reaches STC 43, while double boards, a deeper 6-inch cavity, resilient channel and asymmetric construction reach STC 51. STC belongs to the whole assembly and cannot be replaced by the NRC printed on the batt.

So the direct answer to "is rock wool better than glass wool for sound" is: there is no material-family winner. Rock wool is usually easier to make into higher-density, stiffer, temperature-resistant, non-combustible engineering board — suited to fire separation, equipment wrapping and damp or stiffness-demanding locations. Glass wool is easier to make low-density, supple, easy to cut and low-cost for full-cavity fill — suited to residential and light-steel cavities, duct lining and broadband reverberation control. Compared at the same thickness, similar flow resistance, same backing and facing, the 125 Hz to 4 kHz alpha curves can overlap almost completely.

For low frequencies, any "just add more fiber" claim must stay humble: a 25 mm glass-fiber board at ASTM C423 Type A shows alpha of only 0.04 at 125 Hz, and 0.11 even at 51 mm, despite NRC of 0.70 and 1.00 respectively. Low-frequency traps, staggered double walls, floating or resilient ceilings and membrane or Helmholtz resonators are the real tools; fiber only damps part of the cavity and does not raise the wall's STC by the same margin.

2. Glossary: Keep "Absorbs 0.9" and "Blocks 50 dB" Apart

ASTM standards must not be mixed, and even the standard names must be written correctly.

Outside North America, the equivalent routes are EN ISO 11654 (sound absorption grading), EN ISO 10140 (laboratory sound insulation), ISO 354 (reverberation room) and ISO 717 for rating — cite the standard family applicable to the project's market.

3. Why a High-NRC Fiber Still Fails Low Bass: Frequency, Thickness, Flow Resistance and Air Gap

3.1 Mid- and high-frequency look strong because the test weights them high

NRC's four anchors start at 250 Hz, exactly where most fibrous batts are most comfortable; 63, 125 Hz and the kick drum, cabinet harmonics and traffic rumble are excluded from that average. So NRC is an excellent indoor reverberation sketch but a poor "wall transmission certificate."

Published technical data for a glass-fiber duct liner board are a good counterexample: the same glass-fiber board, ASTM C423 Mounting A (against a solid backing), at 25 mm gives 125 / 250 / 500 / 1k / 2k / 4k Hz alpha of 0.04 / 0.26 / 0.63 / 0.91 / 0.99 / 0.99, NRC 0.70; at 51 mm it gives 0.11 / 0.64 / 1.12 / 1.14 / 1.06 / 1.08, NRC 1.00. Note alpha can exceed 1.0 in the reverberation-room method: it comes from the standard's definition of apparent absorption for finite-sample edges and mounting conditions, not "114 percent of energy absorbed."

Three design rules from the curve:

1. Thickness beats piling on density. For a given formulation, making the material thicker usually moves the absorption peak to lower frequencies better than compressing the same volume harder — but available cavity depth, duct cross-section and ceiling height are hard constraints.
2. 125 Hz is not "70 percent of NRC." It is an order of magnitude below NRC; it needs deep cavities, membrane or panel resonance or a separate mass layer.
3. Facing and mounting change the whole curve. Bare fiber, perforated metal, thin fabric, spray coating and an off-wall air gap are acoustic boundary conditions, not decoration. Copying alpha without copying mounting makes a spec non-executable.

3.2 "Denser is not necessarily more absorptive" is real, not mystic — there is an optimal flow resistance

Porous absorption is roughly three things: the sound wave squeezes into the fiber network, air in the pores does viscous work, and the structure's heat conduction dissipates part of the energy. Too loose, and air passes through with little resistance — little reflection, little dissipation. Too hard and dense, and the wave reflects at the surface before reaching depth. So for a given thickness, fiber recipe and backing, there is an optimal flow-resistance band, not a monotonic "density goes up, absorption goes up."

A parametric study on a fixed 28 mm composite panel found an optimum around 600 cgs over a 50 to 600 cgs flow-resistance scan, and porosity and structure factor shift the peak: optimal porosity differs between low and high frequency bands. This is not a recipe table saying "all glass wool should be 600 cgs"; it is a counterexample showing that treating kg/m3 as the only knob misses fiber diameter, orientation, binder, compression recovery and facing.

Practical use:

• Low-density glass wool (commonly about 10 to 48 kg/m3 flexible batt or board): easily reaches low-to-mid flow resistance, good for deep cavities and broadband reverberation control; crushing it past recovery sacrifices low frequency.
• Mid-density rock wool board (commonly about 40 to 150 kg/m3, acoustic or fire boards higher): stiffer, suited to exposed ceilings, equipment-jacket lining and core behind perforated panels; can still be surface-impedance-limited.
• High-density "soundproofing wool" (above 150 to 200 kg/m3 or more): if it is merely stiffer mineral wool, it remains primarily an absorptive, thermal and fire core and does not automatically become a mass barrier. If a product claims transmission performance, it must supply E90 band TL or assembly STC — not just density.

3.3 The backing air gap is a "free" low-frequency lever, up to a depth

Mounting thin fiber hard against a rigid wall (C423 Mounting A) versus offset from it (Mounting E or a custom air gap) usually improves low frequency, because the air gap adds equivalent acoustic thickness and changes the resonant boundary. The rule is not "deeper air gap is always better": past a point the cavity impedance mismatches and the gain saturates; and if both leaves of a cavity wall are rigidly tied to the same stud, low-frequency isolation is still short-circuited by the stud.

Two layers of thinking:

• Absorption optimization: fiber thickness plus offset distance plus perforation or facing transmission loss together; a recording studio first-reflection point cannot be decided by NRC alone.
• Transmission optimization: first make the two masses as independent as possible. The air gap itself has no surface mass; its value is providing decoupling distance for a double wall or resilient ceiling, not "leave 2 inches of empty space to replace a second board."

4. Product-Level Comparison: Same-Basis Data Only

4.1 Representative published product data (verifiable samples)

The values below are manufacturer technical data sheets or published standards, not this site's lab retests. Acoustic values must be cited with test standard, mounting, sample size and backing; products with different mountings cannot be ranked head-to-head.

4.2 Density: an engineering map, not a material law

The "sweet spot" is an objective function, not a physical constant. For a low-cost 2x6 wall cavity needing reverberation control and thermal performance, soft, well-recovering, uncompressed glass wool is usually easier to get right; for a visible rigid board under a perforated metal ceiling, near a fire zone or in a damp mechanical room, rock wool board's dimensional stability and non-combustible core earn their place. For 80 Hz isolation, both wools step back to a supporting role.

5. Fire and Temperature: Separate Four Things

5.1 Correct the temperature framing

The common claim "glass wool −120 to 400 °C, rock wool ≤650 °C" cannot be a universal answer for North American acoustic selection, for four reasons:

1. The temperature scale may be misaligned. A typical North American HVAC glass-wool lining TDS quotes a service air-temperature limit of 121 °C (250 °F) (ASTM C1071 or C411 context), not 400 °C; a −120 °C low limit must come from the specific binder, facing and frozen-moisture SDS, not assumed for all glass wool.
2. "Glass softening or fiber melting point" is not "bonded-and-faced product serves continuously." Glass-fiber itself softens well above 400 °C, but the thermoset binder, coating, foil or FSK, adhesive and sealant in a duct liner fail first.
3. Rock wool fiber core melts around 1000 °C magnitude, not a guaranteed 650 °C continuous service. Some high-temperature industrial mineral wools exceed or fall below 650 °C continuous; the SKU's SDS "physical or chemical properties – thermal decomposition or conditions to avoid" and the manufacturer temperature curve decide. Writing "≤650 °C" as the rock wool family ceiling simultaneously misses more temperature-resistant special products and wrongly condemns ordinary boards with organic binder or facing.
4. Building fire hours are an assembly property. ASTM E119 gives a wall, floor or protection system its performance and duration under a specific fire exposure; "non-combustible" or E84 Class A does not auto-convert to 1 h or 2 h. The design number, stud spacing, board layers, screws, penetrations and batt location from a recognized testing laboratory must all agree.

5.2 Executable temperature-verification checklist

One-line red line: take the service-temperature limit as the worst of fiber core, binder, facing, adhesive, sealant and airflow erosion; take the fire hour from the whole listed assembly, not the marketing page.

6. Installation and Indoor Air Quality: Not "Carcinogen vs Safe"

6.1 Correct boundary: distinct from asbestos, but cutting dust still controlled

Man-made vitreous fibers (MMVF) — including glass wool, rock or stone wool and slag wool — differ from asbestos in crystalline morphology, persistence and toxicological evidence. For glass wool, IARC moved it from 2B to Group 3 in 2001: "not classifiable as to its carcinogenicity to humans"; a 2011 epidemiological review still found no consistent evidence of increased respiratory-system cancer risk from glass-wool exposure. This is not a "zero risk" or "throw it around bare-handed" pass; Group 3 means the evidence is inadequate to classify.

The real on-site issue is mechanical irritation and migratable fiber or dust: skin itch, eye, nose or throat irritation, transient cough; cutting, sawing, tearing packaging and blowing or disturbing old batts sharply raise exposure. Rock wool is not necessarily "more toxic" and glass wool not necessarily "safer"; it depends on fiber fineness distribution, binder, whether pre-faced, dust load and work method. Conversely, branding all mineral wool as carcinogenic is equally unprofessional scaremongering.

6.2 Contractor-executable exposure control

On indoor air quality, once the product is reliably faced and enclosed, migratable-fiber risk is a different matter from bare construction; but spray binder, VOC sources, moldy wet wool and dust residue must also be managed in the IAQ plan. "Formaldehyde-free wool" does not mean skip facing, vacuuming and cut control.

7. Scenario Selection Matrix: Pick the Failure Mode First, Then the Family

7.1 Cavity fill: correct target, wrong expectation

Correct expectation: damp the parallel gypsum, stud and air system, reduce some cavity resonance and panel-radiation coupling, cut cavity reverberation; on an already high-mass assembly there is often "perceptible but limited" STC gain, frequency-dependent. The assembly table treats wool as a fixed variable while STC moves from 43 to 51 with stud depth, board layers, resilient channel and asymmetry — exactly proving wool is not an independent variable.

Wrong expectation: treating R-13 or R-21, NRC 0.9 or "high density" as an STC increment promise; stuffing wool into a single-board wall blocks bass; one side of 2 in wool replaces a second mass board. If electrical boxes are unsealed, bottom track unsealed, receptacle boxes shared-wall, penetration holes leak, even the most expensive wool only absorbs a little reverberation in a leaking cavity.

7.2 Studio and home theater: low-frequency target must be written as bands

An executable acceptance draft:

1. Write the target spectrum, not just "STC 60" or "NRC 0.8." Example: 63 / 125 / 250 Hz room impulse response or RT60, low-frequency modal decay, background-noise NC or RC, and the wall's field ASTC or FSTC.
2. Use wool where there is depth: corner trap 100 to 200+ mm, rear-wall mid or high first-reflection 25 to 50 mm up, ceiling by reflection path and HVAC noise.
3. With perforated board or transmissive fabric, test face plus wool as a system; perforation, cavity depth and wool thickness together set the peak — you cannot add "board NRC" and "wool NRC" separately.
4. When bass is insufficient, first add equivalent depth or membrane or panel resonance, not another layer of high-density rock wool. If the room mode is 40 to 80 Hz, a few centimeters of wool contributes too little to be the main scheme.

8. From Metrics to Construction Drawing: Six-Step Design and Check

Step 1 — Define the source spectrum and receiver tolerance. Start from the source, not "need soundproofing wool": speech or TV (above 500 Hz matters), subwoofer or band practice (63 to 250 Hz), AHU broadband noise floor, pump or compressor (discrete low-frequency line plus broadband), site boundary traffic (low frequency and impulses). Write target as band sound pressure level, NC or RC, RT60 or field ASTC — not "quiet."

Step 2 — Draw the transmission paths, find the short-circuits first. Red-line every path: direct through-wall, shared ceiling plenum, door gap, cable hole, receptacle box, drain pipe, structural beam, exterior window, floor joist, duct sleeve. On site, confirm the main path with level difference, phase or headphone listening or professional transmission testing. Most failed projects did not pick the wrong wool; the air bypass and solid short-circuit were still there.

Step 3 — Use mass, decoupling and seal to set the main transmission layer. For airborne sound, first estimate the ideal mass-law ceiling: TL_mass is approximately 20 times log10 of (f times m) minus 47, where f is in kHz and m is in kg/m2. It only shows trend: raising effective surface mass from 10 to 20 kg/m2 gains about 6 dB in the mass-controlled region; from 20 to 25 kg/m2 the gain is far less than sealing one 1/8 in gap. Then add board count, staggered studs, resilient channel, independent wall, resilient ceiling or floating slab. Wool's surface mass is usually only about 0.02 to 0.2 kg/m2 magnitude — it cannot stand in for a mass layer.

Step 4 — Use thickness, flow-resistance and air-gap to design absorption. Treat wool as a controllable damper: (1) pick frequency target — mid or high reflection control 25 to 50 mm; below 125 Hz needs 75 to 150+ mm or membrane or panel composite; (2) compare C423 alpha(f) within the same product family, not NRC as the only column; (3) if perforated facing, require the "facing plus wool plus air gap" system curve; if exposed, require the exposed-state curve; (4) keep design thickness, no crushing, no top or corner voids. Acceptance may use reverberation-time before and after, but note furniture, people, doors or windows and HVAC state differences.

Step 5 — Write temperature and fire as reviewable clauses. Temperature clause: "maximum continuous air or surface temperature plus peak plus duration plus medium," not "withstands 400 °C." Fire clause: "E119 or ULC design number plus hours plus permitted substitution table plus penetration fire-stop system," not "non-combustible = 2 h." HVAC also writes velocity limit, erosion, anti-fungal or mold, moisture uptake, corrosion and serviceability.

Step 6 — Site acceptance and review. Installation completeness (sample or borescope or photo of top, behind receptacles, duct bends — no compression, no continuous seam, continuous facing); acoustics (field E336 or ASTC or room acoustics, recording source, doors or windows, HVAC — by project acoustic spec, not manufacturer NRC); transmission weak points (intensity or scan gap, smoke pen or pressure differential — no dominant leak, main path improved to budget); air and hygrothermal (velocity, static or drop, condensation, IAQ cleanup verification — by MEP or building-physics design); health and safety (SDS, PPE, local exhaust, training, post-construction vacuum — by employer, OSHA or provincial authority and site assessment).

9. Specification Verification Checklist (copy into CSI format)

Acoustic clauses

• Submit full ASTM C423 frequency table (from 125 Hz to at least 4 kHz), with mounting (A, E or actual construction), sample size, air gap, facing, reporting lab and date; NRC only as a derived column.
• If supplier gives impedance-tube data, label it explicitly ASTM E1050 normal-incidence alpha; do not mix with C423 NRC or claim "equivalent NRC."
• If claiming wall or ceiling system STC, ASTC, CAC or OITC, submit the corresponding ASTM E90, E336, E1414 or E1332 assembly report; stud size, board thickness or layers, cavity, fill, seal, doors or windows and test lab must match the construction drawing.
• For studio or low-frequency targets, separately submit 63 to 125 Hz system performance or let the designer use an acoustic model; NRC alone is not accepted for low frequency.
• State density unit and test method (kg/m3 or pcf, ASTM C167 etc.); do not present "high density" as a number.

Fire, temperature and durability clauses

• SDS and TDS temperature sections separately: continuous service, peak or short-term, thermal decomposition or shrinkage, unusable conditions; do not substitute fiber melting point for product limit.
• For exposed or duct application, submit ASTM E84 (flame spread and smoke) and applicable UL 181, ASTM C411, C1104, C1338 reports; state whether the airflow-face coating or facing was in the sample.
• Fire separation must cite the ASTM E119 corresponding assembly design number, hours, permitted material-substitution table and penetration or joint fire-stop system; ordinary wool's "non-combustible" is not a substitute basis.
• HVAC clause includes max velocity, erosion, dusting, corrosion, moisture uptake, condensation, microbial control and replaceability.

Construction and IAQ clauses

• Each batch submits SDS (WHMIS or HazCom applicable items), TDS, origin or batch, asbestos-free declaration (especially important for old materials or mixed supply chains).
• Before installation, sample-confirm cut method, facing seam, recovered thickness, contact with duct or stud; no air-blowing cleanup.
• PPE and exposure control per site exposure assessment: long sleeves or pants or gloves, eye protection, local exhaust, respiratory protection per SDS and regulation with fit management.
• Before handover, vacuum, seal holes, check diffuser or lighting or access with no exposed fiber; record whether HVAC startup shows odor, visible fiber plume or abnormal pressure drop.

10. Final Selection Advice: Decide in One Sentence, Accept in One Page

• Lowest cost to cut cavity resonance and improve thermal and acoustic fill in residential or light-steel walls: choose well-recovering, uncompressed, steadily supplied glass wool; spend the saved budget on sealing, board layers and fewer penetrations.
• Near fire separation, mechanical room, exposed acoustic ceiling, damp or stiffness-demanding board: put rock wool system on the shortlist first, but verify the North American SKU's acoustic report, temperature and assembly list.
• Recording-studio mid or high clarity and reverberation control: either works; thickness, off-wall air gap and transmissive facing matter more than the family name; require alpha(f).
• To block bass, neighbor speech or home-theater leakage: first design mass plus decoupling plus sealing, wool only fills the cavity; doors, windows, ducts, receptacle boxes and floor or ceiling bypass each need a separate budget line.
• High temperature or fire hours: stop at every "melting point" marketing page; instead check SDS service temperature, E119 or ULC design number and field substitution clause.

The most important professional discipline: next time you see "NRC 1.0, soundproofing miracle" or "high-density rock wool, STC off the charts," first ask what the ad means by the object — material, facing system, or whole wall? 1 kHz alpha or 125 Hz alpha? Laboratory E90 assembly TL, or a salesman borrowing the reverberation-room number as transmission? Answer those four questions completely, and glass wool vs rock wool selection becomes a verifiable, acceptable, accountable noise-control engineering — not a material superstition.

For material deep-dives, see our guides on glass wool, rock wool and glass wool acoustic insulation, and the broader glass wool vs rock wool: 7 differences comparison.