Convolution Reverb
Convolution reverb is a digital signal processing technique that reproduces the acoustic character of a real physical space by convolving an audio signal with an impulse response (IR) — a recording of how that space responds to a brief broadband transient. The IR captures the full decay signature of the environment, including early reflections, diffusion, and tail, allowing the processor to mathematically superimpose that acoustic fingerprint onto any dry signal. Unlike algorithmic reverbs built from feedback networks and filters, convolution reverb achieves photorealistic spatial accuracy because it is derived from actual acoustic measurements.
Most producers believe that convolution reverb is always more realistic and therefore always the superior choice over algorithmic reverb for professional work.
Convolution reverb is more acoustically accurate, but accuracy is not always the goal — algorithmic reverbs are more musically flexible, easier to modulate, and less CPU-heavy, which is why the world's top mixers use both types simultaneously. Convolution is the realism anchor; algorithmic is the creative sculptor. Additionally, a static IR cannot capture the nonlinear, dynamic behavior of analog hardware, meaning that for some applications (like recreating the warmth of a Plate 140 driven hard), an algorithmic emulation or the actual hardware may outperform the convolution of its own IR.
What Is Convolution Reverb?
Convolution reverb doesn't simulate a room — it teleports your track inside one.Convolution reverb is a digital signal processing technique that reproduces the acoustic character of a real physical space by convolving an audio signal with an impulse response (IR) — a recording of how that space responds to a brief broadband transient. The IR captures the full decay signature of the environment: early reflections, diffusion patterns, modal resonances, and the long tail that defines the room's size and surface character. The processor then mathematically superimposes that acoustic fingerprint onto any dry signal, sample by sample, in real time. The result isn't approximated — it's measured. You are not guessing what a cathedral sounds like; you are playing the actual cathedral back as a filter around your audio.
The critical distinction between convolution reverb and its algorithmic reverb cousins is where the spatial information originates. Algorithmic reverbs construct their sense of space from feedback delay networks, all-pass filters, and modulation — mathematical architectures that approximate room behavior without ever measuring a room. Convolution reverb derives its character from an actual physical measurement. That distinction produces a perceptual gap you can hear immediately when you A/B them: algorithmic reverbs are recognizable as reverb; convolution reverbs are recognizable as places. The difference becomes most stark on acoustic instruments, orchestral arrangements, and vocals where the listener's ear has a finely calibrated sense of what "correct" physical space sounds like.
The impulse response is the engine. To capture one, an engineer fires a starter pistol, pops a balloon, sweeps a sine wave through the audio spectrum, or plays a specially designed chirp signal inside the target space. A measurement microphone captures the result — every reflection, every flutter echo, every millisecond of decay as the room's energy dissipates. That recording is then deconvolved to extract the pure room response, stripping out the character of the excitation source itself. What remains is the acoustic transfer function of the space: a precise mathematical description of how that room transforms any sound that enters it. Load that IR into a convolution processor, and every signal you pass through it will decay exactly as it would have in that room, at that position, with that microphone placement.
What makes this relevant for producers beyond academic interest is the perceptual result. A plate reverb or room reverb algorithm produces a response that sounds reverb-like. A convolution reverb loaded with an IR of Abbey Road's Studio Two produces a response that sounds like Abbey Road's Studio Two. That specificity is what makes convolution the default choice for orchestral mockups, cinematic scoring, jazz records, classical recordings, and any genre where placing an instrument inside a real acoustic environment matters more than designing a new one. The early reflections land in the right place, the diffusion builds at the right rate, and the tail decays with the natural frequency-dependent absorption of actual materials — carpet absorbs highs faster, concrete holds them longer, and the IR preserves that behavior verbatim.
The practical limitation is the inverse of the strength. Because convolution reverb captures a fixed acoustic event, it cannot be reshaped the way an algorithmic reverb can. You can't widen the stereo field the way a plate reverb plugin allows, or morph the density pattern in real time, or dial in a completely unnatural decay shape. What you can do is select a better IR, apply pre-delay, scale the decay length, and EQ the reverb return — and in most mix situations that is enough. Understanding where convolution's precision becomes a constraint is as important as understanding where it excels.
— Bob Clearmountain, Mix Engineer (Bruce Springsteen, The Rolling Stones, Bryan Adams) — Sound On Sound — Classic Tracks: Bryan Adams Run To You, March 2010"Reverb is not decoration. It's the room the music lives in. Get the room wrong and the music feels homeless."
Convolution reverb mathematically imposes the measured acoustic fingerprint of a real space onto any dry audio signal, producing spatial realism no algorithm can replicate from scratch.
How Convolution Reverb Works
The mathematical operation at the core of convolution reverb is called convolution — specifically, linear convolution of two discrete-time signals. Take your dry audio signal as one sequence of samples and the impulse response as a second sequence. Convolution multiplies every sample of the dry signal against every sample of the IR and sums the results at the appropriate time offsets. In practical terms: every single sample of your vocal or guitar is treated as its own tiny impulse, the IR is "stamped" at that position with the appropriate amplitude, and all those stamped copies are summed together across time. The result is that each moment in your dry audio is followed by a decaying copy of the room's acoustic response. That accumulation of overlapping, time-offset room responses is what you hear as reverb.
The computational challenge is why convolution reverb required specialized hardware before it became a software plugin. A one-second IR sampled at 44.1 kHz contains 44,100 samples. Convolving one second of audio against that IR requires roughly two billion multiply-accumulate operations per second. The solution that made real-time convolution possible is the Fast Fourier Transform (FFT). By converting both the audio and the IR into the frequency domain, performing multiplication there instead of in the time domain, and converting back, the processor reduces the operation count from O(N²) to O(N log N) — an enormous saving that modern CPUs handle comfortably. This is called fast convolution or frequency-domain convolution, and it is what every convolution reverb plugin does under the hood. The perceptual result is identical to time-domain convolution; the computational cost is manageable.
One artifact of FFT-based convolution is latency. Partitioned convolution — which breaks the IR into shorter blocks processed sequentially — is the industry solution, allowing low-latency operation at the cost of slightly higher CPU usage. The quality of partitioned convolution implementation separates premium plugins from budget alternatives: poorly partitioned algorithms introduce subtle pre-ringing or timing smear that makes the reverb feel slightly disconnected from the dry signal. When a high-end convolution plugin costs ten times what a stock one does, the difference is often in this implementation detail — the quality of the FFT engine, the partition size strategy, and how the plugin handles the transitional seam between the early-reflection and late-reverb portions of the IR. The gain staging of the IR itself also matters: an IR captured too hot will clip internally in the convolution engine, coloring the decay in ways that register as harshness rather than room character.
Convolution reverb achieves real-time operation by converting audio and IR to the frequency domain, multiplying them there, and converting back — a process that makes sample-accurate room simulation computationally viable.
Convolution Reverb — Key Parameters
Convolution reverb has fewer moving parts than an algorithmic reverb, which means each parameter carries proportionally more weight. There is no density knob, no modulation depth, no diffusion slider — what you have is control over the IR itself and how your signal interacts with it. Master these six controls and you can place any source in any acoustic environment with clinical precision.
The interaction between pre-delay and wet level is where most convolution reverb decisions actually happen. Increasing pre-delay lets you run the reverb return hotter without degrading intelligibility — because the direct sound has already landed and been processed by the listener's ear before the reverb arrives. A 25ms pre-delay might allow 30% wet where 0ms pre-delay at 30% wet would feel washy. This relationship means you should always set your pre-delay before dialing in your wet level; reversing that order produces a setting that needs to be readjusted once you add the delay.
Decay stretch and high-frequency damping interact in a way that rewards careful listening rather than visual parameter matching. A stretched IR at 150% with aggressive high-frequency damping can produce a slow-blooming, intimate-feeling reverb even from a large-hall IR — a texture frequently used on ambient electronic productions and film score pads. The stretched tail gives the sense of size; the damped highs remove the acoustic signature of the original room and replace it with a felt sense of distance rather than a heard sense of space. These are two different perceptions, and convolution reverb lets you dial between them.
Convolution reverb's six core parameters are fewer than an algorithmic reverb offers, but each one controls something acoustically real — making every adjustment carry more perceptual consequence.
Quick Reference Card
Setting at least 20 ms of pre-delay before the convolution reverb onset is the threshold at which the human auditory system perceives the dry vocal as distinct from its reverb tail — below this, the room blurs into the attack of every word and pushes the lead vocal back in the mix. This is the single most actionable number when dialing in convolution reverb on any front-of-mix element.
These are working starting points for convolution reverb in common production contexts — not ceilings. Adjust wet level by ear at matched levels with the mix, never in solo.
| Source | IR Type | Pre-Delay | Decay / RT60 | HP Filter | Notes |
|---|---|---|---|---|---|
| Lead Vocal | Small Hall / Chamber | 20–30ms | 0.8–1.4s | 100Hz | High-shelf cut above 8kHz; run on send at 100% wet |
| Acoustic Guitar | Small Room / Chamber | 8–15ms | 0.5–0.9s | 80Hz | Keep IR size plausible — don't put a fingerpicked guitar in a cathedral |
| Orchestral Strings | Large Hall | 25–40ms | 1.8–2.5s | 60Hz | Decay stretch at 100% for realism; push to 120–130% for epic scoring |
| Snare (Room) | Small Room / Plate | 0–5ms | 0.3–0.6s | 120Hz | Short IR keeps the crack; high-pass aggressive to protect kick |
| Piano | Concert Hall / Chamber | 15–25ms | 1.2–2.0s | 80Hz | Match IR to the piano style — jazz piano rarely needs a 2s hall |
| Choir / BGVs | Cathedral / Large Hall | 30–50ms | 2.0–3.5s | 80Hz | Longer pre-delay separates words from wash; damp highs above 6kHz |
| Synth Pad | Ambience / Short Room | 0–10ms | 0.2–0.5s | 100Hz | Pads already have width — subtle room presence only; avoid cathedral |
| Film Score / Cinematic | Orchestral Hall / Custom IR | 20–35ms | 2.0–4.0s | 60Hz | Use the same IR across all elements for a unified acoustic world |
Tools for This Entry
Signal Chain Position
Convolution reverb sits after compression and saturation in the signal chain for a deliberate acoustic reason: reverb applied before compression means the tail gets compressed along with the source, smearing transients and making the onset of decay audible in an unnatural way — the reverb surges as the compressor pumps. Placing reverb after compression means the dry signal arrives at the convolution processor at a consistent, controlled amplitude, and the IR decays naturally from that point. On a send-return setup — which is the correct architecture for convolution reverb in nearly all contexts — this ordering is automatic: the post-fader send receives an already-processed signal and feeds it to the reverb return. The return itself can then receive additional EQ treatment, notably the high-pass filter and high-frequency shelf cut, before it enters the mix bus. Reverb before EQ on the return is a signal chain error that wastes the precision of the IR's high-frequency behavior.
Interaction Warnings
- Compression before reverb on the same channel: If you are running convolution reverb as a channel insert rather than a send, place any channel compression before the reverb insert, not after. A compressor after a convolution insert will react to the reverb tail, not just the dry signal, producing gain reduction that pumps in time with the decay — an artifact that reads as unnatural movement in the room.
- Low-end accumulation from multiple reverb returns: Every convolution reverb return contributes low-frequency energy to the mix. Running three or four reverb buses simultaneously — a common practice on complex sessions — builds sub-200Hz mud cumulatively. High-pass every reverb return independently; don't assume one shared HP is sufficient. Check your mix bus spectrum analyzer for low-mid buildup when engaging all reverb buses simultaneously.
- Saturation or distortion after convolution: Applying saturation or distortion after convolution reverb — either on the return or on the bus — drives harmonics into the reverb tail, producing an excited, unnatural decay that sounds like a room full of distorted speakers. This can be an intentional creative effect, but it should never happen accidentally. Keep saturation upstream of the reverb send, not downstream of the return.
History of Convolution Reverb
1960s–1970s: Academic Origins and the Measurement Problem
The theoretical foundation of convolution reverb is inseparable from the field of room acoustics measurement. In the late 1960s and early 1970s, acoustic researchers including Manfred Schroeder at Bell Labs were developing mathematical tools for characterizing room impulse responses — primarily to solve architectural problems like concert hall design and speech intelligibility in public spaces. Schroeder's work on pseudo-random noise sequences as measurement stimuli, published in the Journal of the Acoustical Society of America in 1979, established the swept-sine and MLS (Maximum Length Sequence) techniques that are still used to capture IRs today. The idea of using a measured impulse response as a computational filter for audio was a logical extension of this work, but the computational power required to do it in real time with meaningful audio quality simply did not exist in 1975. The gap between knowing the math and being able to use it musically was enormous, and it would take two decades to close.
1980s–Early 1990s: Hardware Precursors and DSP Development
The hardware reverb units of the 1980s — the Lexicon 480L, the EMT 250, the Sony DRE-2000 — were algorithmic, not convolution-based, but their development drove the DSP chip improvements that would eventually make convolution feasible. The first hardware convolution units appeared in specialized academic and broadcast audio contexts in the late 1980s, and by the early 1990s Sony had developed internal prototypes capable of convolving audio with measured IRs, though not yet at commercially viable processing speeds. The Yamaha SREV1, released in 1999, was the first commercially successful hardware convolution reverb unit targeted at professional recording studios — a rack-mount unit with dedicated DSP that could load custom IRs and process stereo audio in real time. Its arrival established convolution reverb as a studio-viable technology and demonstrated that the acoustic realism gap between algorithmic and IR-based processing was audible and commercially meaningful, not just academically interesting.
2000s: Software Arrives and the Workflow Changes
The plugin era began in earnest with the release of Audio Ease Altiverb in 2000, the first commercially successful software convolution reverb, initially released as a Mac AudioSuite plugin for Pro Tools. Altiverb shipped with a library of measured IRs from real spaces — scoring stages, concert halls, cathedrals, living rooms — and its acoustic accuracy immediately made it the default choice for film and television scoring in Hollywood. Logic Pro's Space Designer (introduced in 2004 with Logic Pro 7) brought convolution reverb to every Logic user as a built-in, royalty-free tool, collapsing a technology that had cost thousands of dollars into something a bedroom producer could use on day one. This democratization happened just as the CPU power required for low-latency convolution became available in consumer hardware — a precise convergence that permanently shifted professional practice. By 2005, convolution reverb was the default starting point for orchestral mockup work; algorithmic reverb had been repositioned as the creative or character-driven alternative.
2010s–Present: IR Libraries, Hybrid Processing, and the Streaming Era
The contemporary convolution reverb landscape is defined by two developments: the explosion of commercially available IR libraries and the emergence of hybrid convolution-algorithmic plugins. IR libraries from venues like the Concertgebouw, Musikverein, and Abbey Road Studio Two became available commercially, giving composers access to world-class acoustics without booking the spaces. Simultaneously, plugins like Exponential Audio Phoenix Verb and Zynaptiq Adaptiverb began blending convolution-accuracy with algorithmic flexibility, and Slate Digital's VerbSuite Classics introduced convolution captures of legendary hardware reverb units — merging the IR-measurement approach with hardware emulation. The LUFS-normalized streaming environment has made long reverb tails a deliberate mix decision rather than a default: tails that extend beyond the loudness normalization window contribute to integrated loudness without adding perceived energy, so producers running convolution reverb on streaming-optimized mixes increasingly use shorter IRs with carefully managed wet levels to preserve dynamic range under normalization.
— Trevor Horn, Producer (Frankie Goes to Hollywood, Pet Shop Boys, Seal) — Sound On Sound — Trevor Horn: The Art of Production, January 2010"Production is about creating a world the listener inhabits for three and a half minutes. Every element — the compression, the reverb, the arrangement — is part of that world's architecture."
Convolution reverb evolved from acoustic research measurement tools into a studio-standard plugin in under a decade, permanently redefining the benchmark for spatial realism in recorded music.
How Producers Use Convolution Reverb
The correct workflow architecture for convolution reverb is the send-return bus, not a channel insert — and this isn't preference, it's physics. Route a post-fader send from your dry channel to a dedicated reverb bus, set the convolution plugin on that bus to 100% wet, and control the reverb level with the bus fader. This approach gives you three critical capabilities: you can blend the reverb level in real time without touching the plugin, you can send multiple sources to the same reverb bus so they share an acoustic environment (essential for making elements sound co-located in the same room), and you can treat the reverb return independently with EQ and compression without affecting the dry signal. For orchestral mockups, run all your string, brass, and woodwind channels to the same hall reverb bus — the shared acoustic space is what makes the ensemble sound like it's in one room rather than assembled from twenty separate sessions in twenty different spaces.
IR selection is where most producers underinvest time. The common mistake is treating IR categories as interchangeable within a category — loading any "hall" IR and assuming it's appropriate for the source. In practice, a German Romantic concert hall IR sounds radically different from a British Victorian hall IR at the same RT60, because the early reflection pattern, the room geometry, and the materials are different. Spend time auditioning IRs from your library against the actual source, playing the mix at a nominal level, not soloing the reverb. The correct IR reveals itself in context: the reverb becomes part of the sound, not audible as an addition to it. If you can clearly perceive "reverb is happening," either the IR is wrong for the source, the wet level is too high, or the pre-delay is too short. The sign of a correctly chosen, correctly set convolution reverb is that bypassing it makes the source feel suddenly naked and unconvincing — not that the reverb sounds beautiful in isolation.
1. Create an Audio or Instrument track with your dry source. 2. Create a new Audio track — rename it 'CONV VERB SEND'. 3. Set that track's 'Audio From' to your source track, set Monitor to 'In', and set the track's Audio To to your Mix Bus or Master. 4. Drop Ableton's 'Convolution Reverb' (Live 11 Suite) or 'Convolution Reverb Pro' onto this return track. 5. Set the plugin's Dry/Wet to 100% Wet. 6. On the source track, set 'Audio To' to the CONV VERB track at the send amount you want. Alternatively: Right-click any return track → rename it → drop Convolution Reverb on it → enable the send knob on your source channel and dial in depth. 7. On the plugin, click the IR slot to browse — navigate to Built-in Library or load your own .wav IR file. 8. Set Pre-Delay to 15–25 ms for vocals. 9. On the return track, insert an EQ Eight after the reverb and apply a high-pass at 100 Hz to clean the low end. 10. Use the return track fader to set final reverb level in the mix.
1. Open Logic Pro and load your session. 2. Create an Aux channel strip in the Mixer (Option+click in the Sends area of any channel to create a new Bus/Aux automatically). 3. On the newly created Aux, insert Space Designer (Logic's built-in convolution reverb) in Insert Slot 1. 4. In Space Designer, set the Output to 100% Wet (drag the Dry slider to zero, Wet to maximum). 5. Click the IR Sample button and navigate to a space — Logic includes an extensive built-in library under Spaces > Halls, Rooms, Plates, etc., or click 'Load IR' to import a custom .wav file. 6. Use the Envelope section to adjust Attack and Release on the IR tail for tonal shaping. 7. Enable 'Reverse' mode for ambient/atmospheric reverse reverb effects. 8. On the Aux, add a Channel EQ after Space Designer and cut below 100 Hz. 9. On your source channel strip, pull up the Send slot, select the Bus feeding this Aux, and set the send level. 10. Engage Pre-Fader on the send if you want reverb independent of channel fader position.
1. In FL Studio 21, open the Mixer (F9). 2. Route your instrument or sample to a dedicated Mixer track (e.g., Insert 1). 3. Select a separate Mixer track for the reverb return (e.g., Insert 10) — route Insert 1 to Insert 10 by enabling the routing button in the Insert 10 send matrix at the bottom of the mixer. 4. On Insert 10, click an empty FX slot and add Fruity Convolver (FL's native convolution reverb) from the effect list. 5. In Fruity Convolver, click the waveform display and load an IR file in .wav format (drag from your browser or use the folder button). 6. Set the Wet knob to maximum and Dry to zero on the plugin. 7. Adjust Pre-delay in the plugin settings and resize the IR using the size/trim handles in the waveform view. 8. Add Fruity Parametric EQ 2 as the next slot on Insert 10 and apply a high-pass at 100 Hz. 9. Control depth using the Send knob from Insert 1 to Insert 10 in the routing matrix. 10. Balance the return track's main fader for final mix depth.
1. Open your session in Pro Tools. 2. In the Mix window, create a new stereo Aux Input track (Track > New > Stereo Aux Input). 3. Set the Aux Input's input to a free internal bus (e.g., Bus 7-8). 4. On your source channel (Audio or Instrument track), pull down the first send slot and assign it to Bus 7-8 stereo. 5. Insert your convolution reverb plugin on the Aux Input — Avid's AIR Reverb, Waves IR-1, or Altiverb 7 are common choices. Set the plugin's Dry/Wet to 100% Wet. 6. Load an IR in the plugin browser — AIR Reverb includes built-in IR presets; Altiverb includes its own library accessed from the plugin GUI. 7. Set Pre-delay inside the plugin to 20 ms for vocal sources. 8. Add a 7-band EQ (or Avid EQ3) after the convolution plugin on the same Aux Input and apply a 100 Hz high-pass. 9. Optionally, insert a bus compressor after the EQ to lightly control reverb tail dynamics. 10. Automate the send level from the source track to ride reverb depth across the arrangement.
When listening to determine whether your convolution reverb is working, focus on three moments: the attack, the gap between words or notes, and the tail decay. At the attack, the pre-delay should be long enough that the source lands cleanly before the reverb onset — if you hear smearing at the attack, increase pre-delay by 5ms increments until the onset clears. In the gaps between notes or words, the reverb decay should fill the space naturally without surging or abruptly cutting — surging indicates the wet level is too high or the IR is too long for the tempo; abrupt cutting usually means the IR was incorrectly stretched below 60% and has a truncated tail. At the decay, listen for the frequency-dependent behavior: the high frequencies should fade faster than the lows, and the tail should feel like it belongs to the same acoustic category as the source. A harsh or metallic tail usually means the IR itself was captured with too-close microphone placement, or the IR has uncorrected modal resonances in its decay — audible as pitched hum in the reverb trail.
Convolution reverb and delay are complementary spatial tools that should be deployed together on most mixes. Delay creates width and repetition that positions a sound laterally; reverb places it in depth — front-to-back positioning in the acoustic field. Running a short slapback or quarter-note delay before the reverb send (or as a separate send feeding a pre-reverb stage) creates a richer, more complex early-reflection structure than any single IR provides. This technique is standard on cinematic vocal production: the delay establishes rhythmic interest and lateral spread; the convolution reverb adds the sense of physical inhabitation in a real space. Together they build a three-dimensional position for the voice that no single effect achieves alone.
The professional workflow for convolution reverb is always send-return at 100% wet, with IR selection happening in mix context — not in solo — until the reverb becomes acoustically invisible.
Convolution Reverb by Genre
Convolution reverb usage varies dramatically across genres, not because the tool changes but because the implied acoustic environment changes. Classical, jazz, film scoring, and acoustic singer-songwriter production are categories where acoustic realism is a core aesthetic goal — convolution is the default. Hip-hop, electronic, and pop production frequently use convolution alongside algorithmic reverb, with the convolution IR serving as a grounding "room presence" while algorithmic reverbs provide the creative, unnatural spatial character. Metal and hard rock production often avoids convolution entirely in favor of parallel compression and tight room algorithms that preserve transient aggression.
| Genre | Ratio | Attack | Release | Threshold | Notes |
|---|---|---|---|---|---|
| Trap | N/A | N/A | N/A | N/A | Short plate or room IR (0.3–0.8 s) on snares and hi-hats; minimal reverb on 808s (tight room IR only); pre-delay 10–15 ms; reverb return high-passed at 150 Hz to keep sub clean. |
| Hip-Hop | N/A | N/A | N/A | N/A | EMT 140 plate IR on vocal chains; medium room IR (0.8–1.5 s) on sampled instruments; pre-delay 20–30 ms on lead rap vocals; reverb return lightly compressed to tame tail dynamics. |
| House | N/A | N/A | N/A | N/A | Club or warehouse IR (1.0–2.0 s) on pads and synth leads; very small or no convolution reverb on kick and bass; gated or short room IR on claps; pre-delay synced to 16th note at session tempo. |
| Rock | N/A | N/A | N/A | N/A | Live studio room IR (0.6–1.2 s) on drum overhead send for natural kit glue; plate IR on snare; medium hall IR on lead guitar solos; vocal gets tight room IR (0.5–0.8 s) to preserve presence in busy arrangements. |
| Mastering | N/A | N/A | N/A | N/A | Convolution reverb is rarely applied at mastering; when used, a very subtle stereo room IR at extremely low wet levels (−30 to −40 dB return) can add micro-ambience and cohesion to an otherwise flat or over-compressed master. |
When the table conflicts with what your ears tell you, trust your ears — but understand why the deviation is intentional. Using a large hall IR on a hip-hop vocal isn't a genre mistake if the production brief calls for it; it is a mistake if you haven't asked why and haven't confirmed it serves the track. The table represents center-of-gravity usage, not absolute rules.
Hardware vs Plugin vs Stock
The functional gap between dedicated hardware convolution units and software plugins has nearly closed in the past decade, but the difference in workflow, flexibility, and CPU behavior remains real. Hardware units like the Yamaha SREV1 or the TC Electronic System 6000 were designed for fixed-installation use in post-production suites and broadcast facilities — their IR libraries were curated, their I/O was analog, and their latency was optimized for hardware-level DSP. Software plugins running the same IRs on modern CPUs match or exceed them in acoustic quality, but the hardware units offered something software rarely does: dedicated processing that didn't compete with the rest of the session for CPU cycles, and analog signal path coloration from their I/O stages. The practical reality today is that a well-configured plugin on a modern machine is the correct choice for 95% of studio scenarios; hardware convolution units are workflow investments for facilities where dedicated processing bandwidth matters more than flexibility.
| Aspect | Hardware | Plugin |
|---|---|---|
| IR Quality | Curated factory libraries, often venue-specific | Unlimited third-party IR import; access to commercial libraries |
| CPU Load | Zero host CPU — dedicated DSP | Moderate to high, depending on IR length and partition size |
| Latency | Fixed, hardware-optimized, typically <5ms | Buffer-size dependent; plugin delay compensation required |
| Flexibility | Limited to built-in IR library and controls | Full parameter control, custom IR loading, automation |
| Workflow | External routing, fixed installation | Fully in-the-box, recallable, DAW-integrated |
| Cost | $3,000–$15,000+ for professional units | $0 (stock) to $400 for top-tier plugins |
Before and After
The dry piano recording sounds close and studio-isolated — presence without dimension, every note existing in an acoustically dead vacuum that exposes the sample library's artificiality and disconnects the instrument from any believable physical environment.
The same piano sits inside a small recital hall IR: early reflections arrive 18 ms after each note placing the instrument at a perceived distance, the 1.4-second tail reinforces harmonic resonances naturally, and the entire performance now inhabits a specific, credible physical space that makes the mockup feel like a genuine recording session.
When auditioning the before and after of a correctly applied convolution reverb, the critical listening point is not the reverb tail — it's what happens to the dry source. The source should feel more three-dimensional after the reverb is engaged, not just wetter. If the before sounds like a recording and the after sounds like a performance in a space, the convolution is working. If the after just sounds like "a recording with reverb on it," something is wrong — the IR is mismatched, the pre-delay is too short, or the wet level is too high. Bypass at matched levels, and ask whether the source feels more real or more processed. The correct answer is always more real.
Convolution Reverb In The Wild
These tracks were selected specifically because the convolution reverb is doing architectural work — defining the physical world of the recording — rather than acting as an audible effect. Listen for the relationship between early reflections and the decay tail, how the space changes the perception of distance, and whether you can hear the specific character of the IR's source material in the way the high frequencies die out differently from the lows.
Across these seven tracks, the consistent lesson is that the most effective convolution reverb applications are the ones you can feel but can barely hear. Radiohead's string arrangement, Hans Zimmer's orchestra, and Max Richter's quartet all use convolution to establish physical location before they use it to create atmosphere. The technique in each case is the same: select an IR matched to the source's acoustic plausibility, set pre-delay to preserve intelligibility, and push the wet level just far enough that bypassing the reverb removes something real from the sound — not just a texture, but a sense of place.
Types of Convolution Reverb
See the full comparison: Plate Reverb
See the full comparison: Room Reverb
Not all IRs are created from the same source, and not all convolution implementations process them the same way. The type of IR you load — and the convolution architecture that processes it — fundamentally shapes the acoustic character of the result. Understanding these distinctions lets you choose the right tool before you start auditioning, rather than browsing presets until something sounds close.
Choosing the correct IR type before loading the plugin is the decision that determines whether convolution reverb sounds like acoustic truth or like a missed tool selection.
The single biggest mistake producers make with convolution reverb is treating IR selection as a preset browse rather than an acoustic casting decision. You are not choosing a texture — you are choosing a room, and that room needs to make physical sense for the source inside it. A dry close-miked vocal belongs in a believable human-scale space; putting it in a cathedral IR and turning down the wet to compensate doesn't fix the mismatch, it just makes it quieter. The diagnostic is brutal: bypass the reverb at matched levels. If bypassing sounds more honest than engaged, you have the wrong IR, wrong pre-delay, or wrong wet level — probably all three. Convolution reverb should make the source feel more true, not more processed.
Convolution reverb is not a reverb plugin — it is a room selection tool, and every mix decision that follows depends on whether you selected the right room.
Common Mistakes with Convolution Reverb
Convolution reverb is acoustically precise in a way that amplifies mistakes rather than hiding them. An algorithmic reverb set incorrectly sounds vague; a convolution reverb set incorrectly sounds physically wrong — like a recording that violates the laws of acoustics. These six mistakes account for the overwhelming majority of convolution reverb errors in amateur and intermediate production work, and each one has a specific, testable fix.
Placing a close-miked vocal inside a large cathedral IR and reducing wet level to compensate doesn't solve the problem — it creates a quiet acoustic contradiction. The early reflections of a cathedral IR arrive 80–120ms after the direct sound, which is physically consistent with being 20–30 meters from the nearest wall. A human ear in a recording environment processes that as fundamentally wrong for a voice at conversational distance. Match the IR's physical scale to the implied recording distance of the source: small room or chamber IRs for close-miked sources, hall IRs for sources that are supposed to feel distant or grandiose.
Running a vocal through convolution reverb with no pre-delay fuses the reverb onset with the consonant attacks of the lyric, smearing intelligibility and making individual words indistinct. The attack of a word and its room reflection arrive simultaneously, and the listener's auditory system can't separate them as effectively as it can when the direct sound arrives first. Set pre-delay to a minimum of 15ms on any vocal reverb — 20–30ms is more common in professional practice. This also makes the reverb feel more natural: in any room, the direct sound reaches your ear before the first reflection, because the direct path is always shorter.
Every convolution reverb return without a high-pass filter is slowly filling your mix with low-frequency mud. A 2-second hall IR applied to a piano carries 2 seconds of decaying low-end energy into the mix for every note the piano plays. Multiply that by a string section, a choir, and a vocal on separate reverb buses, and the sub-200Hz region of your mix becomes acoustically congested in a way that no amount of kick or bass EQ can fix, because the reverb mud is time-smeared across the entire bar. Default to a high-pass on every reverb return at 80–100Hz and push it higher — up to 150Hz — on busy or bass-heavy arrangements.
Choosing an IR based on how it sounds with the reverb return soloed — at 100% wet, no dry signal, listening to the pure room response — produces settings that fall apart the moment you unmute the mix. The reverb's job is to make the source feel placed in a space when heard in context, not to sound beautiful in isolation. Always audition IRs with the full mix playing at nominal listening level. The IR that sounds most impressive soloed is almost never the one that sounds most natural in the mix. The correct IR is the one you stop noticing.
Stretching an IR to 150% or 200% of its original length to get a longer tail introduces time-domain artifacts that the convolution algorithm cannot conceal. The natural frequency-dependent decay of the room — where highs fade faster than lows — becomes mathematically distorted at extreme stretch values, producing a tail that sounds increasingly synthetic and tonally skewed. If you need more tail than the IR provides at 100%, find a longer IR in the same acoustic category rather than stretching the existing one. Decay stretch is a subtle adjustment tool, not a room size control: stay between 80–120% for acoustic realism; treat values beyond 130% as intentional creative distortion.
When two or more sources need to share an acoustic environment — a lead vocal and a guitar in the same room, or all the instruments in an orchestral arrangement in the same hall — running convolution as an insert on each channel rather than routing them to a shared reverb bus means each source gets its own copy of the room. The resulting reverb is technically identical but perceptually separate: the room responses don't interact, accumulate, or cross-excite the way they do in a real shared space. Use a single send-return bus for any group of sources that should feel co-located, and control each source's contribution to the shared room via its individual send level.
Every convolution reverb mistake has the same diagnostic: bypass the reverb at matched levels and ask whether the source sounds more true engaged or disengaged — and if the answer is disengaged, find the specific parameter that is creating the acoustic contradiction.
Red Flags and Green Flags
Red Flags
- 🔴 Reverb tail builds up low-end mud below 150 Hz — you haven't high-passed your convolution reverb return, causing the IR's low-frequency content to compete with kick, bass, and subs.
- 🔴 Zero pre-delay on lead vocals — the reverb onset smears the attack of every consonant, destroying lyric intelligibility and pushing the vocal back in the mix.
- 🔴 Using the same large hall IR on every element in the mix — the result is a one-dimensional soup where nothing has its own spatial identity and the mix loses depth and separation.
Green Flags
- 🟢 Pre-delay set between 15–30 ms on vocal and instrument IRs, giving the dry transient space to breathe before the room envelope begins — a hallmark of professional spatial placement.
- 🟢 Different IR sizes assigned to different elements (tight room on drums, medium hall on guitars, large hall on strings) — creating a believable three-dimensional depth field across the mix.
- 🟢 Convolution reverb blended with a short algorithmic pre-reverb on the same send, combining the realism of the IR tail with the smearable, shapeable attack of an algorithmic unit.
Red flags in convolution reverb applications almost always point back to one of three root causes: the wrong IR for the source's implied acoustic environment, insufficient pre-delay creating onset smear, or uncontrolled low-frequency accumulation from multiple reverb returns. When you see these flags in a mastering session — a muddy, undefined low-mid, or a mix that sounds reverberant but unclear — the fix is almost never on the master bus. It's in the individual reverb return high-pass settings and pre-delay values that weren't set correctly during the mix. Green flags are simpler: the source feels placed, not processed; the reverb is only noticed when it's bypassed; and the space feels physically consistent with the recording's implied acoustic environment. When convolution reverb is working, you don't talk about the reverb — you talk about where the song lives.
Your Progression with Convolution Reverb
Convolution reverb has a shallow learning curve to basic competence and a steep one to mastery. Getting a hall IR to make a piano sound bigger takes twenty minutes. Understanding why a specific IR from a specific position in a specific room is the correct choice for a specific source in a specific mix — and being able to execute that choice consistently — takes years of deliberate listening practice. The three stages below describe what changes at each level and what specific skills unlock the next one.
Load your DAW's built-in convolution reverb (Ableton's Convolution Reverb, Logic's Space Designer, or a free IR loader), select a hall or room IR preset, drop it on a send bus, set wet to 100%, and dial in the return fader until the source instrument feels placed in the space — not swimming in it. At this stage, the goal is understanding the difference between insert and send routing, and developing the habit of always adding a high-pass filter on the reverb return before you commit to a level. These two practices alone put you ahead of most beginners.
Learn to match IR selection to source acoustic plausibility, not just sonic preference. Build a curated library of ten to fifteen IRs across four categories — small room, chamber, hall, and plate — from verified high-quality sources such as the free Fokke van Saane IR library, Logic's Space Designer factory content, or the Voxengo Impulse Modeler. Practice setting pre-delay by ear: start at zero, increase in 5ms steps while the mix plays, and stop when the source onset clears without the reverb feeling disconnected. At this stage, also learn to automate the reverb return level through sections — opening in the chorus, contracting in the verse — so the space breathes with the automation of the arrangement.
Advanced convolution practice means building a unified acoustic world across an entire mix — selecting a primary hall IR used consistently across all instruments in the session, supplemented by secondary ambience IRs for individual close presence. Capture your own IRs using a measurement microphone, a sine sweep, and free software such as Room EQ Wizard to record the acoustic signature of physical spaces you have access to. Experiment with hybrid convolution-algorithmic processing by routing the output of a convolution reverb into a short algorithmic tail processor, or by using mid-side processing on the reverb return to widen the side content of the tail while keeping the center signal tight. At this level, you should be able to identify the IR category used on a major record by ear and replicate its spatial character from scratch.
Progression with convolution reverb is a progression in acoustic literacy — the ability to hear what a space should sound like and select, configure, and integrate the IR that delivers it.
Frequently Asked Questions
An impulse response is a recording of how a physical space or hardware device reacts to a near-instantaneous broadband sound — typically a starter pistol shot, a sine sweep, or an electrical impulse. You can download free IRs from sites like OpenAIR or Fokke van Saane's collection, or purchase commercial IR packs from companies like Waves, Altiverb, or Ircam; you can also record your own with a measurement microphone and a sine sweep.
Convolution reverb uses a mathematically exact fingerprint of a real space and is prized for acoustic realism — it sounds like the actual place. Algorithmic reverb uses networks of delays, filters, and feedback to synthesize a plausible room, making it more flexible, CPU-lighter, and creatively malleable but less physically accurate. In professional mixing, the two are often combined: convolution for realism, algorithmic for sculpted or textured effects.
Almost always as a send/return configuration with the plugin set to 100% wet. This lets you share one reverb instance across multiple tracks (saving CPU), control the dry/wet balance with the return fader, process the reverb return independently with EQ and compression, and automate the send depth without affecting the dry signal's level.
The most common cause is low-frequency energy in the IR tail accumulating in the 80–250 Hz range — apply a high-pass filter at 80–120 Hz on the reverb return bus. Additional culprits include too much reverb on low-frequency instruments like bass or kick drum, and too short a pre-delay that smears transient attacks into the reverb onset.
First, increase your audio buffer size during the mixing stage. Second, use your DAW's freeze or commit function to render the reverb to audio. Third, truncate your IR to the shortest useful length — much of the CPU cost scales with IR duration, so cutting the tail after the perceptually useful decay (often after -60 dB) can halve processing load without audible difference.
Yes — hardware IRs are a major category of impulse response. Many engineers have measured the EMT 140, EMT 250, Lexicon 480L, AMS RMX16, and other classic hardware units, and these IRs are widely available commercially and free. The convolution of a hardware IR captures the tonal and temporal character of the physical unit very accurately, though some engineers argue that the nonlinear, dynamic behavior of analog circuits is not fully preserved in a static IR.
A standard starting point is 15–30 ms of pre-delay on lead vocals, which creates a perceptual gap between the dry vocal attack and the reverb onset — this preserves intelligibility and keeps the vocal seated up front in the mix. For backing vocals or doubled parts you can reduce this to 8–15 ms. Pre-delay values above 50 ms start to become audible as a discrete slap effect, which can be creative but typically pulls focus from the lead.
No — 'better' depends entirely on the creative goal. Convolution excels at acoustic realism, natural instrument spaces, film scoring, and period-correct hardware emulation. Algorithmic reverbs are superior for creative manipulation — you can adjust density, modulate the tail, apply gated effects, or dial in a fantasy space that doesn't exist in nature. Many professional engineers describe their workflow as using convolution to anchor instruments in a believable world and algorithmic processing to add texture, glue, or movement on top.