What is the Haas effect?

The Haas effect in plain English

In any room, most of what reaches your ears is late. First comes the sound that travelled straight from the source, then copies of it that bounced off the floor, the walls and the ceiling, all within a few tens of milliseconds. If your brain treated each of those as a separate event, every word anyone said would come with a cloud of echoes. It groups everything that arrives inside a short window into one sound instead, and takes the direction from whichever copy got there first.

That grouping is the Haas effect, and it works the same way with two speakers. If a guitar plays from both and one is 10 ms late, you hear one guitar, placed at the speaker that was on time. Because the late copy is still there at full level, the guitar also seems to take up more room than it did.

Producers use it to widen mono parts. Put the part on both sides, delay one side by somewhere between 5 and 30 ms, and it spreads out. You pay for it in mono, where the two sides get added back together and the delay turns into a comb filter.

Haas, Wallach and the precedence effect

Helmut Haas studied the effect for his doctoral thesis at Göttingen, and his paper on how a single echo affects speech came out in 1951. He played speech from two loudspeakers, delayed one of them, and asked listeners when the second became a disturbing echo. For delays of roughly 10 to 30 ms, the late loudspeaker could be up to about 10 dB louder than the first before most people objected, and 10 dB is roughly twice as loud.

Others got there around the same time. Hans Wallach and his colleagues named it the precedence effect in a 1949 paper, and Lothar Cremer had already described a law of the first wavefront. All three names mean the same thing; hearing scientists mostly say precedence effect, and studios mostly say Haas.

Delay time and what you hear

Play the same sound from two speakers with one side late, and it goes through three stages as the delay grows. The borders between them are soft and move with the material, but the order never changes.

Under 1 ms: the sound moves

Your ears are about 18 cm apart, so a sound from one side reaches the near ear up to about 0.6 to 0.7 ms before the far one. Your brain uses that gap, along with differences in level, to work out where a sound is. Delay one speaker by a fraction of a millisecond and you create the same kind of gap, so the sound slides towards the speaker that's on time, and by about 1 ms it has gone most of the way. This is called summing localisation: the two speakers still add up to a single image, and the delay just moves it.

From about 1 to 30 ms: one wider sound

Past a millisecond or so the image stops moving. It stays at the earlier speaker however much delay you add, but the late copy changes how it feels, and the sound gets bigger and less pinned to one point. This is the range people mean when they say Haas delay. Most of the useful settings sit between 5 and 30 ms, since below 5 ms the result is mostly a lean to one side.

Sound covers about 34 cm in a millisecond, which is a handy number to keep in your head. A 12 ms delay is the same as moving one speaker about 4 metres further away from you.

Past the echo threshold: two sounds

At some point the late copy stops fusing and you hear it as its own event. That point is called the echo threshold, and it depends on the sound more than on anything else. For a single click it can be as short as 5 ms, and for speech and sustained music it can reach 30 to 50 ms, with drums somewhere in between depending on how sharp the hits are. Level counts too, because a late copy that's nearly as loud as the first one is easier to pick out than a quiet one.

Keep going and you get a clear echo. From about 80 to 150 ms it's a slapback, which has its own lesson in slapback delay.

Reading the lab

The lab plays a loop through a stereo signal path where the right channel can arrive late. On this page only Delay R, Pan and the Mono button are live; the other controls belong to the phase lesson and are greyed out here.

The goniometer, the round screen at the top left, plots left against right. When both sides play the same thing the dots stack into a vertical line. Delay one side and they spread into a fuzzy ball, because at any instant the two speakers are playing slightly different moments of the waveform. The correlation meter puts a number on that: +1 means the sides are identical, and the reading falls as they become less alike. A Haas delay can take it below zero, even though nothing has been flipped.

The wide screen headed What mono does to the tone sends a click through your settings and shows what happens to the tone with left and right added together. With no delay it's a flat line. Add a delay and it becomes a row of dips, labelled with their frequencies; hover over it, or press and drag on a phone, to read any point.

Delay R goes up to 50 ms: the Haas range is covered in the first 30, and the rest lets you hear where a sound stops fusing. The Right now panel explains what your settings are doing to the sound you picked, using numbers the lab measures from that sound as it plays through the same path you hear. Start with the nylon guitar: it was recorded with one microphone, so anything you hear moving was caused by the delay.

How to widen a sound with the Haas effect

Any mono part with most of its energy above the bass works: a strummed or picked guitar, a synth line, a shaker, a backing vocal. This method works in any DAW.

1. Get the part onto both sidesDuplicate the track and pan the original hard left and the copy hard right. To keep it on one track, use a plug-in that can delay one channel, such as Logic's Sample Delay, which has separate left and right settings.
2. Delay one side by 10 to 15 msAbleton Live has a Track Delay box in each track's mixer section that takes milliseconds, and most other DAWs have a similar offset per track. A plain delay plug-in on the copy, set to one repeat with no feedback and 100% wet, does the same job.
3. Sweep the time by earShorter times lean more to one side. Longer ones sound bigger, until the part starts to flam or sounds like a second take. Percussive parts usually end up shorter than smooth ones.
4. Balance the leanThe part will sit towards the side that plays first. Turn that side down a dB or two, or nudge the pan, until it sits where you want it.
5. Check it in monoSum the mix to mono and listen to the part. If it goes thin or phasey, try a different delay time, turn the late side down, or take the low end out of the late side with a high-pass filter.

Balancing the lean

A Haas delay always pulls the part towards the side that plays first, and on a busy mix that can leave one side heavier. Level and time trade against each other. Make the late side louder and the image creeps back towards the middle, though Haas's experiment shows how much it takes before the late side wins: a late loudspeaker up to 10 dB louder was still heard as part of the first sound. In a mix you'll rarely want more than a few dB between the sides.

The tidier fix is to balance one Haas part against another. If one guitar is early on the left, set the next one up early on the right; they lean in opposite directions and the mix stays even.

The Haas effect in mono

A Haas delay works on your ears because the two copies arrive from different places. In mono they come out of the same speaker, so they add up as waveforms before they reach you, and wherever the delay puts a frequency exactly half a cycle out of step, the two copies cancel. That happens at a row of evenly spaced frequencies, which is a comb filter.

The first notch lands at 1 ÷ (2 × delay) and the rest follow every 1 ÷ delay above it. At 12 ms the first notch is near 42 Hz, the next at 125 Hz, then 208 Hz, and so on every 83 Hz. A 1 ms delay puts the first notch at 500 Hz with one every 1 kHz above it, so there are fewer notches and each is much wider, which sounds boxy. Longer delays pack the notches closer together, and the part goes hollow and phasey instead.

Mono playback is more common than it seems. Phone speakers, a lot of Bluetooth speakers, many club and venue systems and anyone listening from the far side of a room all hear something close to a mono sum, which is why the advice in the mono compatibility lesson applies to every Haas delay you set up.

You can make it less severe. With the late side turned down 6 dB, the notches are about 10 dB deep instead of complete, and that's usually much harder to hear. A high-pass filter on the late side keeps the low end out of the comb altogether, which is where cancelling does the most damage, since the bass carries most of a mix's weight.

Haas effect or panning?

Both move a sound towards one speaker. Panning does it with level: each speaker plays the same waveform at the same moment, so nothing can cancel in mono and the part simply returns to the middle. A Haas delay does it with time, which also makes the part sound wider, and the time difference is what causes the comb filter in mono.

Pan when you only want to place something. Use a Haas delay when a mono part sounds too narrow and you've checked that it survives the mono sum. The two are set side by side in Haas effect vs panning.

The Haas effect outside the mix

Every room relies on it. The early reflections off nearby surfaces arrive inside the Haas window, so they colour the sound and tell you about the room without coming across as echoes. The early reflections lesson covers that side of it, and the reason a little pre-delay keeps a vocal clear of its reverb is in the pre-delay lesson.

Live sound uses it on purpose. At a large gig, speakers further back in the crowd are delayed so that the sound from the stage reaches you first. Engineers usually add a little extra on top of the exact distance, anywhere from a few milliseconds to about 20, so the main system always wins the race and you hear the band from the stage, even when the delay speaker next to you is doing most of the work.

Spaced pairs of microphones rely on time differences too. With two mics a metre or so apart in front of a choir, a singer on the left reaches the left mic first, and that small gap is part of how the recording places them.

Common Haas effect mistakes

Putting it on the low end

A Haas delay on a bass or kick puts the first mono notch in the bass range and takes the weight out of the mix on any mono system. Keep low-end parts in the centre and widen something else.

Never checking mono

A Haas delay sounds its best where you're most likely to judge it, on headphones or between two speakers. Flip the mix to mono each time you add one.

Using it on everything

Each Haas part you add pulls more of the mix out of the centre and stacks up more comb filtering in mono. A couple of widened parts can make a mix feel wide, while a dozen leave it with no clear middle and a hollow mono version.

Going past the echo threshold

On a percussive part, 25 or 30 ms can come across as a flam or a sloppy double. If the part starts to sound like two takes, bring the time down until it fuses again.

Widening the lead vocal

The lead vocal usually anchors the middle of a mix. Put a Haas delay on it and it drifts off-centre and changes tone in mono, so widen the backing vocals and leave the lead alone.

Train your ears

The quiz under the lab plays a loop with hidden settings and asks whether it will survive being played in mono. In the rounds that fail, the right side is a millisecond or so late or flipped upside down; in the others the width may change, which mono ignores. The screens are covered, so you have to judge by what you hear between your speakers or in your headphones. Eight rounds take a few minutes, and your best score is saved in your browser.

Questions people ask

Keep going