but no rosemary, parsley nor sage.

Something I just read by Learjeff which seems to be very true and worth remembering…


Anything under 15 milliseconds of latency should be OK, and under 7 or so is almost unnoticeable (unless you’re pretty darn sensitive to it). Many early MIDI synths didn’t respond much faster than this.

reminded me of something I recently read; perhaps you can add it to your Fasoft FAQ’s Fish. :)

0 - 5 mS. For the average human, this time interval doesn’t exist. Two sounds with this spacing sound as one. A sound with a duration in this period is barely heard, no matter how loud. At best, there’s just a “what was that?”, or “did you hear something?” response.

5 - 25 mS. Sounds with this duration are heard, but are very rare in nature. Again, two sounds lying this period apart are heard as one, but the tonal quality of the first sound is altered by the second.

This is the area early reflections are heard. But, they are not heard as separate echoes nor reverberation, rather they just effect the tonal quality of the original sound.

This is also the area where the Haas Effect comes into play.

The Haas Effect: Imagine you’re sitting in front of a stereo pair of speakers, and a equal amplitude monophonic signal is being fed to each speaker.

The sound will appear to emanate from between the speakers, from the middle in fact.

Now, delay the signal going to the RH speaker by 5-25mS, and the sound will now appear to be coming from the LH speaker only.

Even if we increase the level of the sound going to the RH speaker, the sound still appears to be coming solely from the LH speaker. It’s only when the RH level is about 6dB above the LH level that the sound starts to move away from the LH speaker. And it needs to be raised by about 10dB before it sits once more in the centre of the soundstage.

25-30 mS: It’s in this range that we start to hear a distinction between two sounds.

30-50 mS: (apparently a boring part of the time chart! :D)

50-75 mS: Duration of a well muted kick drum or tight snare with no reverb. A good PPM will be fully responding by this time. A VU meter will barely be twitching.

100 mS: a typical staccato note. VU meter starting to make a noticeable response.

250 mS: Typical quarter note (metronome on 60 BPM). VU meter fully responding.

500-750 mS: Typical snare with moderate reverb.

1000 mS: time it takes Toke to write 20 political posts! :laugh:


Hey Ali,

You are just a veritable fountain of knowledge! Very interesting indeed. One other thing about time…it just keeps going…and going…and goin…and…


1000 mS: time it takes Toke to write 20 political posts!
I can beat that!!! :laugh:
I can beat that!!!


It's not my knowledge TG, I got it all from a book; Live Sound Reinforcement by Scott Hunter Stark.

(well, all except the bit about Toke, I worked that out myself).

It goes into great detail about psychoacoustics, not just time, but frequency, directivity, etc., etc.

For example, did you know that the ear is a frequency analyzer?

When a complex sound enters the ear, all the frequency components are sent separately to the brain, which then integrates them all together again. But, people can be, have been, trained to bypass that integration and "hear" the separate partials.

I never realized how important the Haas effect is for sound reinforcement. If you have speakers at the back of a church for example, not only do you have to include a delay to allow for the time sound takes to travel from the front to the back, but you have to include an extra delay so that the sound appears to be coming from the front.

If I have time, I'll write up some of the other psychoacoustics from the book.


Excellent stuff! …But I’m sure Toker is more capable than that. :O

Give us more!


I’m pretty sure Stark is mistaken about some of this, or perhaps an editing mistake on your part.

Based on the standard HRTF models, and assuming the diameter of your head is roughly 6", 0 to 90 degree delay is in the range of 0 to 600 microseconds. The formula is:

ITD=B/2c(theta + sin(theta))

- B is the distance between the observers ears.
- c is the speed of sound
- ITD is the delay to produce the effective direction theta. (I forgot what IT stand for, but D is for delay.)

This is an approximation. A closer approximation includes the distance to the sound, but over about 1.5 feet, the effect of this term diminishes to a neglible amount.

Trust me that if you have a 50 msec delay, it will feel REALLY bad playing a piano – if you’re used to playing real instruments at all. It throws my timing off, bad as my timing is anyway. At 25 msec, it’s a bit annoying but I can live with it, but I doubt a serious player with precise timing would.

When listening to music, a difference of 25 msec isn’t uncommon. However, with very tight, excellent musicians, greater accuracy is required. Excellent musicians can achieve superb effects by coordinating their timing. For example, New Grass Revival used to get a VERY big sound with only guitar, banjo, filddle or mando, and bass – and often this big sound when playing only one or two notes, not big chords, on most instruments. The big sound comes from having the notes come together precisely to sound like one big instrument rather than 4. This agrees with what Ali says above, of course.

One might interpret it to mean that intervals under 25 msec aren’t important, and that’s not the case. Frnakly, it’s the difference between “loose” and “tight”.

BTW, I have tested this and it works. I don’t remember where I found a delay plugin where I could plug in delays in microseconds, but I did.

The strange thing to me was how it sounded exactly like panning. I could have the same volume going into both ears, but one delayed, and it would sound just like panning.

The other interesting thing is that the effect was stronger than panning. For example, I could pan 60 degrees to the right using the delay, and then pan 60 degrees to the left using the stereo pan control, and the result was still to the right of center. (Don’t remember how much, though.)

Try it yourself and see what you find. Experiments should always be reproduced! I need to do my own over again, frankly, in case I’ve misremembered some of the results, and to avoid the placebo effect.

My earlier test wasn’t scientific; I set the controls and listened. For a scientific test, I’d find a means to set the controls randomly, and for each setting, record my response, and then go back and correlate the control settings to the results. So, the results might have been colored by expectations. But since I was surprised by the results, I doubt it.

I’m not quite with you here Jeff, are you talking about the Haas effect?

And, what piano?

And I thought I’d made it clear that 50 mS is in the range where we do hear an echo, not one sound.

I’ve not mis-edited, I’ve triple checked that, but whether Stark is right or wrong, I could not say:

However, if you follow these links:


That one gives the Haas effect as working up to about 25-35 mS.

and here

gives it as up to 50 mS.

And several other links all confirm the validity, and the approximate delay time, of the Haas Effect as I quoted it.

But, they do differ somewhat. The upper limit of the Haas effect is quoted by some as 50 mS, and by others, as low as 20 mS.

The lower limit varies even more, from under 1 mS, to about 6 or 7 mS.

I suspect, (but don’t actually know), that Stark’s figures give an average, after all, different people have different responses, and even the same individual can vary during his life, and during his day.

So, as a general, good enuff, rule of thumb, 5 mS seems reasonable as a lower limit. After all, the book in question is about practical sound reinforcement, and that must cater to half deaf old grannies as well as razor eared kids and sound engineers.

But, perhaps I’ve misunderstood what you’re talking about Jeff, (probably because I’m still trying to figure where the piano fits in :D )

And perhaps I have mis-edited by leaving out some of the finer details.

Yes, I agree with the results of your equation. Sound travels at roughly 1 foot per millisecond, so yeah, from one ear to the other takes about 600 microseconds.

But, that equation neglects the response time of the brain/ear before a transient is registered, and that can take several milliseconds.

(However, looking at your equation, (which I don’t know if it’s right or not), I assume that theta is dimensionless, otherwise the RHS of the equation has the dimensions of time/angle, not time.

So, if the angle is dimensionless, it must be in radians, and I’d be interested in hearing how that is justified).

Anyway, as I said, I’m either misunderstanding what you’re saying, or I never said what I was saying clear enough (and knowing me, it’s probably both :D ).


250 mS: Typical quarter note (metronome on 60 BPM). VU meter fully responding.

Sorry to be a nitpicker, but 250ms is a sixteenth note at 60 bpm (1sec/quarternote).

Someone had posted some months ago about a delay plugin that incorporates the Haas effect. Does anybody remember any details about that plug?

I wonder if latency works differently in different contexts. Mostly you guys are talking about delays in a signal to a listener. Learjeff’s comment also referred to latency in keyboards. The Minimoog had a latency of about 10 ms. I get as low as 3 ms on my computer when running some soft synths in stand alone mode (e.g., try some of the free ones from Big Tick). I can easily tell the difference in responsiveness from 15 - 10- 5 ms. 10 ms is workable, one tends to play ahead of the note a bit. 3 ms is great. You can feel the difference between 10 and 3 really easily. But isn’t this a different issue than the stuff concerning the haas effect, or delays used as comb filters, or the like? This is about how it feels to the player, without worrying about other effects of which delayed signals. All of which matter, of course! :)

keep going though, this delay/panning stuff is great. Really get’s me thinking about some ways to mix.

yesterday, coincidentally, I tried playing with some slight delays because I noticed when I combined my S/PDIF with my Mic inputs from my Vetta II (four channels of the same signal) it turned from a 4x12 cabinet sound to a 6" speaker sound.
So I played around with some different delay amounts all under 5ms to see how you could combine those in a way to make one track seem larger. I noticed the panning issue and thought I would experiment with that as well.

This could be a great way to pan into a crazy stereo image :)

Guitar69, you are clearly hearing the effects of comb filtering as a result of different latencies. If you pan the two signals you will lose the comb filtering to some extent, and additionally the sound will change drastically when summed to mono. I try to keep comb filtering effects panned exactly the same, to avoid summing to mono problems. But such filtering is, as you know, a killer technique for expanding (or contracting) guitar sounds. :)

This is all very interesting to me. I have annual audiograms on the job and they always say “You’re fine.” Well what I would like to KNOW is WHY in an even slightly noisy place, if you want me to understand what you are saying to me, one has to SPEAK UP and look me right in the face. Otherwise, I interpret “My, you look nice.” as “Would you like some rice?” Does my brain/ears have problems dealing with all the delays/echoes in an environment such as a restaraunt or banquet hall?

I’m serious. I wanna KNOW why my hearing works this way. Heck, might keep me outta the doghouse sometime…any ideas?


I have annual audiograms on the job and they always say "You're fine."
It all depends on what they are looking for and how extensive the test is. Your broadband volume perception is probably fine. The first thing that goes is frequency response. You might have a dip in the mids or highs. Another likely thing that hearing loss can cause is sensitivity to frequencies. And then there is the inability to distinguish sounds with there is a lot going on. Depending on the level background noise this is normal, but it can be a sign of hearing loss even if everything else seems fine. There will be other signs if there is loss and they look deep enough.
Sorry to be a nitpicker, but 250ms is a sixteenth note at 60 bpm (1sec/quarternote).

You're perfectly right, and I hadn't noticed that I must confess. It's one of those silly mistakes.

Well, if nothing else, it shows that there's at least one inaccuracy in the book. :D


The last time, they said “You have a slight loss in the higher frequency. But it’s normal for your age.” (41) I don’t know. It’s just annoying and my wife says she does’nt have the problem, “You are just ignoring me!” Of course, THATS impossible!

Another example, we are at home, normal household racket going on, wife or child walks into the room and starts talking. Fine so far, then while they are talking they turn away to walk back in the other room. The last bits, I perceive as a garbled mess. Maybe I need to see a specialist. I don’t know what they could do. I suppose if some of it’s gone…it’s gone…


It’s not only frequency loss, but discrimination loss too.

I have the same problem. I can hear a pin drop in an empty room, but in a noisy environment, I cannot discriminate the person’s voice who’s speaking to from other nearby conversations.

And like you TG, I need the person to be facing me to understand them.

But HF loss also means a loss of directivity in hearing too, so that may have something to do with it, insomuch as we’re unable to “home in” on specific directions as much as we used to be able to.

But it’s #### irritating to be accused of “selective deafness”, I know that much.

Yes, it is selective deafness, but not by choice.


Discrimination loss. That’s an excellent description. At night, outdoors, I hear bugs and other critters making noises nobody else seems to hear. Put me in a “busy” room though…“Er…whats that?..I did’nt catch that one…” and "Please, LOOK at me when you speak?"

You’re right “selective hearing” as my wife calls it. “I SWEAR I don’t do it on purpose honey!” Irritating indeed.


first thing that goes (for men) is the 4khz frequencies. I had a test ten years ago and they monitored a 6 db drop at 4k and a 3 db drop around 3.4k.

I shudder to think what it would be like now.