"My existing speakers sound thin when I turn down the volume - can Harbeth help me?"

[A.S.]

We had a call to the office today asking this extremely interesting question. The full question was something like this ....

'I have UK-made speakers from a well known brand. They sound great when I play them loud, but when I turn down the volume, for example to listen at night, there is no bass. They sound thin, cold and uninvolving. Surely all speakers are the same, but I really want warmth and involvement; could a Harbeth solve my problem? How and why?'

Great question, and one at the very core of the Harbeth sound. I'm a little busy right now but I'll answer this (one of my favourite questions because it's really at the core of the Harbeth differentiation) - this evening!

In the meantime. he's a clue ... here

[hifi_dave]

Can I guess what the 'existing' speakers are ?

[A.S.]

You probably could guess - there would be plenty of choice: the horrible 'modern sound' is bright, thin, reedy, cold, analytical, soulless, peaky, harsh and 'pushed' onto the listener. But the public - who attend very few acoustic concerts - have bit by bit, reprogrammed themselves into thinking that that is the norm. Of course, as they don't hear live unamplified music, why would they think otherwise?

OK, to answer the question. Where do we start? I can't design PA or rock and roll speakers because I have no interest in them and I don't know enough about how they would be used (or misused) and what expectations their users have of them. But in the case of a Harbeth, I have a very clear understanding of how our users will use our speakers, how loud they will play them, how far away they'll sit and what type of music; and above all, what would sound right to that user in that room, with that music at that listening level.

How can you be so sure that Harbeth's designer can read the user's mind? Some psychic powers perhaps?

No! I know how you'll use a Harbeth because I have you, in your room right at the front of my mind when I design because you and I listen the same way, the same distance, the same loudness. The sort of music you'd like I would too.

Does Harbeth design for a big room? Or perhaps for a UK-size listening room?

Harbeth's R&D centre is a cottage deep in the woods, not a building on an industrial estate. I do not design for big rooms because I don't live in a big house, all the rooms are multi-functional and have to double-up as bedrooms, lounges, studies or whatever. And The Cottage, the main listening room, is a perfectly normal domestic environment. So the rooms I have to listen in and work in is on the smaller side of ideal, and the acoustics are real-world acoustics, not million-dollar acoustically perfect environments. Those foam lined, quasi-anechoic expensive purpose-built listening rooms with text book acoustics can seriously delude the designer. Not one customer will listen in that or similar environment.

Does that mean that the Harbeth design/listening room different from other speaker brands?

Yes, I believe it does. Most speaker companies have large, purpose-built (and very impressive) listening rooms which they proudly invite journalists to. And that's where the manufacturers design review panel would sit as a group, X-factor-like, judging the loudspeakers presented to them. But, the problem is that if you design for a large damped room, by definition the speakers are far away, and by definition played quite loud, those speaker will be fundamentally unsuitable for a smaller room.

Why wouldn't they sound right in a smaller room?

Because, by definition, you would sit nearer to them and play them quieter in a smaller room. Those two factors have disastrous acoustic consequences.

Harbeth was founded by a BBC engineer -- why is that significant?

If you understand the importance of Harbeth's BBC foundation you'll really understand the Harbeth sound. The BBC is a broadcast organisation and produces thousands of hours of speech every year. All of that speech has to be carefully monitored and edited and made to fit the available broadcast time. This means trimming needless words and even the intake and exhalation of breath so that what we hear at home, even if heavily edited, sounds natural. In a multi-studio organisation like the BBC there are few studios that are dedicated to just one type of programme so that any one studio in any given week may be used for talks, drama, music recording, voice overs, trailers; the monitor speakers must be capable of revealing difficulties with the source material that allows the studio manager (sound engineer in BBC speak) to make adjustments at his mixing desk for computer workstation. Now the problem in a multi-studio complex is that in one studio there may be a quartet being recorded for an arts programme whilst next door -- through the relatively thin wall -- an obituary may be being read, the next studio along a serious political interrogation and next door to that a pop music programme being prepped.

The uniqueness of the Harbeth monitor concept is that it recognizes that to the sound editing staff in all four of these studios are obliged to monitor at a far lower level than they would ideally like. Now contrast with music recording studios, which are often located in remote and quiet areas in the countryside, where one artist will book the studio and work on his recording in isolation. There is no concern about neighbours and no worry about his performance in the studio or on playback monitors in the control room bleeding through to other artists because there aren't any. This is a wholly different situation to that which a studio manager would find in a broadcast organisation. In the recording studio the monitoring level will be as high or higher than on the artists side of the glass (which is why so many commercial recording engineers have significant hearing damage). Conversely in a broadcast organisation the monitoring level is far below that considered normal in the recording world, may be lower than on the artists side of the glass and is about the level that a hi-fi listener would experience at home.

A feature of the BBC monitor therefore has been to design it such that it has a full warm bodied sound when used at a low to moderate listening. It's optimised for sounding right in the 80 to 100dB range. Conversely a loudspeaker designed for recording studio use and/or one which was optimised for use in a large listening room far from the listener would sound right in the deafening (even frightening) 100 to 120dB range. But swap these speakers over and things don't sound right at all: the BBC monitor will sound rather rich and the high-level optimised speaker will sound very bass shy.

It's all about psychoacoustics. That's a fancy way of saying that the ear is very non-linear and perceived bass in a strange, but completely predictable way that is linked to how LOUD the sound is. This has been extensively researched for well over fifty years and it's very well understood. Read ISO-226

In short then, Harbeth speakers are designed to sound right at home at a normal level because they take into account the psychoacoustics of the ear. I have no idea what psychoacoustic model other speaker philosophies use, but I'd be surprised if their designers had given much thought to the reality of the ear as opposed to their test equipment. Why not phone their sales office and ask what hearing model underpins their design? If they can't point to ISO-226 or the earlier Equal Loudness Contours then I'd say that there is no possibility that their speakers could sound natural at home at a moderate listening level. Try it!

[yeecn]

We had a call to the office today asking this extremely interesting question. The full question was something like this ....

In the meantime. he's a clue ... here

From the graph - the bass region would need a varying of boast to make it sounds 'right'. I would think that that the boast would have been done by the performing artists (e.g. the conductor adjusting the number of instruments to balance the sound scape), with some fine tuning by the recording/mixing engineers. So shouldn't the speaker response be as flat as possible?

But then what sounds flat at 40dB would have too much bass boast at 60dB. Furthermore if the sound engineer does the mixing while monitoring at 80dB, the bass would sounds thin when listened to at 40dB. What if the monitoring speakers at the mixing studio is not flat? I believe that how much bass is right, and what sounds right altogether, is pretty much a personal preference. I can imagine many rock & roll fan will not agree to Harbeth sound. The ear will get used to the irregularities of the sound reproduction system sooner or later. Music is an emotional response, and goes much deeper than mere sounds and individual notes. Beethoven was almost completely deaf by the time he composed his famous 5th symphony!

This is not to say that sound system is not important. While a world class musician can 'hear' the whole symphony by reading the scores, mere mortals like will require a very good sound system just to be able to tell the clarinet and oboe apart, not to mention hearing the counter melodies played by the second violins. I used to hate the soprano voice - it sounded horrible in low-fi system, and it took me 30 years to change my perceptions, thanks to the sweet Harbeth speakers.

?Music works upon our nervous system, our feeling of life, and creates a symphony out of our emotion. In other words, a summary of feeling is somehow gestured out of us through the impact of music, whereas a note-by-note analysis of it would have no meaning. You cannot somehow find out how music has that emotional effect. It is more magical, more arbitrary, more free than that.?

―Adi Da Samraj

[yeecn]

What I meant to say is that - being 'flat' is dB dependent. If I understand the graph correctly, the lines at the bass regions are more lifted and compressed than the mid region, i.e. lower dB requires a higher base boast for it to sound 'flat'; and higher dB requires comparatively less amount of bass lift. This means that if we take one dB level as being 'flat', then increasing the dB level across the frequency range would mean our ear would hear a bass boast, conversely decreasing the dB level would mean we will hear a dip in the bass response.

Now there is the performers playing at a certain dB level, and the recording engineer recording it at another level (probably at lower dB level than the performance level), each probably trying to make it as 'flat' as he/she perceives. But then how it would sound will depends on the listening volume again.

There are so many variables in between. It is probably meaningless to talk about a flat frequency response, at least not for the human ears.

The Audyssey equalization system is meant to address this problem. This system is meant to apply different amount of boast to the highs and the lows dynamically according to the playback volume. My humble Denon 1909 receiver is endowed with such a system. I experimented with it somewhat and found the bass too heavy for my liking. It could be my room acoustics, or it could just be my personal preference. I don't think there is a definitive answer to it.

[A.S.]

Sounding 'right' in the lower frequencies has nothing whatever to do with the recording process where the recording engineers, as you correctly say, are really obliged to record 'flat' (let's assume that anyway) since they don't know what level their recording will be played back at*. The issue is not in the recording, but the reproduction which means the speakers and the room.

So, would you like to take your discussion to the next step?! It may well answer your own question as to why the Denon sounded bass heavy when the 'adjuster circuit' was turned on.


*Or maybe they do. For example, if you were recording for playback in, say, an aircraft on headphones at a low level with lots of ambient noise, the recording engineer would surely be wise to optimise the recording for that application alone. It would sound very wrong if then played back on hi-fi speakers at a high level and without the roar of the aricraft interior.

[yeecn]

I like to clarify that the bass boost by Denon is on the subwoofer - not on the Harbeth speakers. I believe that the boost in the subwoofer is to compensate for the tiny surround speakers, which I think is a deficiency in the Audyssey system. It should give more weights to the front and center speakers. Optimally the surround speakers require their own subwoofers.

In fact I have not been bothered about the Audyssey system lately as I find it makes very little noticeable difference to the Harbeth speakers. The cutover frequency to the subwoofer is set to 40Hz (auto detected by Audyssey). For the type of music that I listened to (classical music) there is hardly any signals going to the subwoofer anyway. This is actually a very interesting point, now that I started thinking about it. I remembered on my old speakers the Audyssey system did make a very marked difference. There was a marked expansion on the frequency range, both high and low.

How the Audyssey system works is that I first have to connect a calibration microphone to the Denon receiver. The microphone is placed in the listening positions (up to six positions). The receiver will then send calibration signals to each of the speakers in turn. From the signals received by the calibration microphone the system will set an equalization value on the various frequency bands to compensate for deficiencies in the speakers and the room acoustic. Now apparently the Audyssey is not doing a lot of compensation to the Harbeth speakers, which means that Harbeth is already more or less optimized for my listening level, which is generally quiet listening.

Now the interesting question. Why some speakers need a lot of adjustments from Audyssey and some don?t. I think it goes beyond how ?good? or ?bad? a speaker is. Given the non-linear nature of human hearing ? if a speaker sounds right at one listening level, it will sound lean and thin when the volume is turned down. Conversely it will appear to have too much bass when the volume is turned up.

From my experience Harbeth is an absolute delight for quiet listening. But it is going to cause too much bass boost on higher listening level? How to design a speaker that will sound right at all listening level? Is that possible?

Now I am finding the Audyssey system very interesting....

p.s.

Audyssey is meant to compensate for the non-linear nature of human hearing as well by applying different level of equalization at different frequency band according to the the playback volume.

[yeecn]

I took some pictures of my setup last night. It is not an example of good acoustic engineering, but some of you may find it amusing to see the extreme conditions that Harbeth speakers could be subjected to.

Picture 1 and picture 2

As you can see, the placement of the speakers is far too close to the walls. But the wife-acceptance factor prevails, and there is not really enough rooms in the living space to move the speakers out anyway. The acoustic of the room is quite bad ? marble floor, cement ceiling, and lots of glass surfaces. The left-speaker is tugged in a tight space, while the left speaker is in a wide opening space between the lounge and the dining space. The spacing between FR and FL is also too far apart.

The center speaker is a pair of P3ES2. The placement is off-center wrt the front speakers, and the cavity behind the speakers does create quite a bit of bass boost. But as I said, wife-acceptance factor prevails.

Picture 3

The surround speakers are tugged behind the small entrance hall. It is an echo chamber, acoustically horrible. But fortunately I don?t watch that much DVD, and I have not come across any music DVD that makes sensible use of the surround speakers yet. I have more or less given up in these speakers for now. I have experimented bringing the speakers forward to where the telephone table is. It sounded much better, but I just couldn't be bothered to redo the wiring now.

Picture 4

This is looking towards the dinning space. You can see a corridor on the right leading to the toilet. That space can create a hell of a BOOM. It happens only occasionally with some sound tracks, but when it happens it was loud. It is a distance from the listening area, so it is not affecting my listening much. I am more concerned about disturbing my neighbours at the back.

As you can see the acoustics of my environment is very compromised. The silver lining is that my wife is considering getting rid of the bulky shelf at the front, and replaces it with a rack of bookshelf with a mini TV console. That would improve the acoustics somewhat. But I am not really looking for perfection. What I have now is good enough for my music enjoyment.

I will follow up with another post on Audyssey calibrations.

[yeecn]

Picture 1

This is the calibration microphone. Well, Denon 1909 was selling for around USD 700. It is very low price in the world of audiophile, so it is questionable how much one can trust the microphone ? and the Audyssey system altogether. The calibration is presented in the screen in 1 octave frequency band. I am certain that Audyssey is doing 2 or 3 sampling per octave on the professional version of Audyssey, but I am not sure with my entry level version.

Remember that the chart corresponds to equalizer setting - i.e. a lift in the chart means Audyssey detected a weakness in the particular frequency band.

Picture 2

Channel level as determined by Audyssey. I am not sure how to interpret the figures. I don?t think it is a constant offset ? but some sort of scaling that is proportional to the playback level. Anyway I find it interesting that FL is 1dB higher than FR. I presume that is because the microphone is picking up more reflections for FR. You can forget about the surround speakers (SL & SR). SR is placed too high, and I think the microphone is not picking up a lot of the higher frequency signals. I need to redo the calibration again with both speakers at the same horizontal level.

Picture 3

Equalizer setting for FL and FR. I find it interesting that there is a definite dip in the bass region, i.e. Audyssey determines that the speakers are giving too much bass, and decided to tone it down. I don?t think this is due to the placement of the speakers alone ? for I do remember that my old speakers gave a completely different curve. It was a typical U curve with a lot of lift in the bass and the high regions for my old speakers. It also tallies with my perception that Harbeth is much more generous with bass as compared with my old speakers.

I also find it interesting that FL got a much higher bass boost as compared to FR. I presumed that is because the curtains and the couch is absorbing a bit of the bass frequencies.

Figure 4

Equalizer setting for the center speaker. The same suppression of the bass region again. The rather sharp lift in the1k band is hmm?..

Figure 5

Equalizer setting for the Surround speakers. The cutover frequency (to subwoofer) is detected to be 125 Hz, so you can ignore the first two bands. Notice that the 250Hz band is lifted off the chart. These two speakers are really weak in bass.

---------------------------------------------------------------------------------------
I find the Audyssey highly interesting, as theoretically it could solve the teething problem of the non-linear frequency response of the human ears, and makes corrections to the speakers and room acoustics in one full scope.

But alas it is of limited use for me now because the subwoofer is calibrated too high to compensate for the deficiency of the surround speakers.

[A.S.]

OK, Yeecn, this Audyssey equalisation system is interesting but it is not exactly the answer we're looking for in this thread. There are, as you have indicated, many electronic solutions for equalising the room and/or speakers in the room to sound good or right or whatever you want. But only a tiny minority of our customers do or would use such a system. No, what we're concentrating on here is how we design Harbeth speakers so that, out of the carton, unaided by electronics, they sound full and natural even at a moderate listening level. This is not an accident of design; it is absolutely deliberate.

So, ignoring the electronics etc. and just concentrating on the speakers and the room how can they work together harmoniously? Thoughts?

[yeecn]

Referring to http://www.harbeth.co.uk/usergroup/s...=4038#poststop, thanks for the very details explanation on this intriguing subject.

I have been pondering a lot about this subject lately. My experience is that Harbeth seems to defy the equal loudness contour graph. I generally listen at fairly wide listening level ? from piano sonata for Yoga at midnight to Beethoven 9th for Saturday noon alone at home. But despite the wide range of listening level, Harbeth seems to always give the same satisfying full range experience.

Now according to the equal loudness contour graph, if the bass is ?right? as low listening level, then the bass would be too loud at higher listening level if the loudness is increased proportionately across the frequency spectrum. But this has not been my experience. It?s been puzzling to me how Harbeth managed achieved a ?balanced? sound for such a wide range of sound level.

Then in one of those late nights while I was lying on the floor I noticed the bass coming from the bottom of the speaker. It was quite substantial, and I realized why Sam insisted that I use his open frame speaker stand. I have since been taking interest of the sounds coming out of the sides, top, bottom, back panels as well as the pothole.

My feeling is that the speaker box, which acts a resonance box, is not linear. I believe that its efficiency decreases as the sound level increases. This gives the necessary depression to the bass response for higher sound level. It would be interesting to attach an accelerometer to the side panels and measure it response at different sound level.

I read somewhere that the speaker cone movement has to increase 4 times for each octave decrease in frequency. There is no way the speaker cone can generate enough movement to create the sound pressure out of low viscosity air. The bass response of the speaker is literally magic created out of thin air. Speaker making is more than science - I believe the non-science part, the magical part, is where the secret lies. With most great creations, scientific investigations is after the fact, and it can never fully account for the creation.

Any, thanks for creating such an endearing piece of art. It?s been an endless enjoyment and fascination for me.