Maximizing Expected Performance of a Workteam: Implications for Population Representation

Suppose we are tasked with assembling a workteam, and we can select for the workteam from two populations, say A and B.  Suppose the expected performance of a person on a workteam can be reduced to a single value v between 0 and 10 (10 is best, 0 is worst) and suppose that value can be reliably and accurately determined through the selection process. And suppose the expected performance of a workteam equals the sum of the expected performance of individual team members.  Thus, for example, if I have a two person team and person 1 has value v=7, and person 2 has value v=8, then the total value of this team is 7+8=15.

Case I: Two populations identical in size and with identical characteristics

Suppose in assembling the workteam we can select from population A with 6 members, or population B with 6 members.

Suppose population A has members with values (10, 9, 9, 8, 8, 7).  We identify the first member of population A as A[1], and thus the value of A[1] here is 10, the value of A[2] is 9, and so on.

Suppose population B also has members with values (10, 9, 9, 8, 8, 7) – so population B is identical to population A.

What observations can we make?

Observation 1: When the supply of the highest value individuals within any given population is too little to staff the workteam(s), then assembling a team from a single population will be sub-optimal.  In this example, to assemble a 2 person team, hiring from only A or only B will be suboptimal. Why? The two individuals with the highest performance values do not both belong to either A or B.  If I choose the two highest valued members of A, the expected performance of the workteam will be 19, and same for B.  However by drawing on A and B the expected performance of the workteam will be 20.

Observation 2: If I want to assemble a workteam of 4 people, the highest performing teams must contain A[1] and B[1], and the remaining two members can come from either A or B as follows:

Team: A[1], A[2], A[3], B[1], expected performance 38

Team: A[1], A[2], B[1], B[2], expected performance 38

Team: A[1], B[1], B[2], B[3] expected performance 38

Note that in 2 out of the 3 optimal solutions, the highest performing workteams will not be composed of equal members from A and B.

Case II: Two identical populations exist, however twice as many candidates from population B are available for selection.

In this case, the effective population size for say A is half that of B. Let:

A = (10, 9, 9, 8, 8, 7)

B= (10, 10, 9, 9, 9, 9, 8, 8, 7, 7)

Notice that B has twice as many members as A, and for every member of A with value v there are 2 members in B with the same value. Thus if I want the highest performing team of say size 3, I will pick 1 person from A (the one with value 10), and two persons from B (also with values 10).  If I want the highest performing team of size 9, I will exhaust the available highly valued candidates from A after picking 3 people from A, so the other 6 will come from B. This leads to:

Observation 3: In general, the optimal team will be composed of 1/3 members of A, and 2/3 members of B simply because there are twice as many available candidates in B, even though A and B can be assumed to otherwise be identical populations. If I were to force an optimal workteam to be composed of 1/2 A and 1/2 B, I would have a sub-optimal solution because there would not be enough of the best candidates available in population A.

Case III: Two populations identical in size and with identical average characteristics but with different distribution of performance values

Suppose now population A has members with values (10, 8, 8, 8, 7, 7).

Suppose population B has members with values (9, 9, 9, 9, 6, 6).

A has characteristics: average value 8, max 10, min 7

B has characteristics: average value 8, max 9, min 6

Observation 4: For this example, the optimal two person team is

A[1], B[x], where x=1,2,3 or 4

Observation 5: For this example, the optimal four person team is

A[1] and 3 members of (B[1], B[2], B[3], B[4])

Notice that even though the single most valued individual is in population A, and even though population A is identical in size and average value to population B, in this example the optimal workteam will be composed of 25% population A and 75% population B.

 

THE Show Newport Beach 2014 – Immediate and Obvious

I’m going to rant and ramble a bit here.  So, this past Saturday I attended THE Show in Newport Beach, and actually I heard quite a few things I liked and had a good time.  It didn’t start so auspiciously.  At first I went to hear the big Focal speakers, the big big ones, the Grand Utopias.  And they didn’t sound good to me.  The bass didn’t sound natural, the highs were overly bright, and my son said, “Dad, nothing sounds good to you.”  What can I say?

This blog post is mostly about speakers, and that’s because in my opinion they make the most immediate and obvious difference in a system.  Yes, maybe amplifier A sounds a bit different from amplifier B, but if you haven’t heard an entire system before, the easiest thing to try and deduce is the sound of the speakers.

Certain things in audio are immediate and obvious.  If a speaker plays loudly cleanly, that is obvious.  If it has good bass, obvious (subject to the program material of course).  So dynamics, and bass extension are immediate and obvious.  With a good recording and an excellent audio system, the sound can be “startlingly real” – and if or when you encounter that, it is immediate and obvious.

What’s my point? In my opinion, most things at an audio show are not immediate and obvious.  What is the difference an amplifier makes, or a turntable – if you are unfamiliar with the whole system, how can you tell?  Getting on my soapbox for a minute, the component I am most skeptical about, especially in the context of an audio show, is audio cables.  Now, if someone has a speaker cable or an interconnect that deliberately changes the signal – e.g., high capacitance for example, maybe there’s a basis for a non-subtle difference.  Otherwise, I think one would be hard pressed to tell the difference among cables under the best of conditions, and I think it virtually impossible under show conditions.  So it is funny to me to enter room after room with big mention being made of the audio cables.  Understand the situation – you walk into say a hotel room, there is a room full of strange (as in not familiar) equipment, the room acoustics are unknown, the listening position is likely too close, you don’t know the recording, you don’t know the cartridge, or the dac, or the amplifier, or the speaker.  And then, given all these unknowns, somehow you are supposed to attribute some of what you hear to the cabling??? Get real. 

After hearing the big Focals, we waited in line, and we waited, probably over a half hour.  What were we waiting for?  It was the MBL Home theater demonstration.  Again, at first, disappointment.  There was an unnatural quality to male vocals, a certain chestiness and lack of clarity.  But the overall sound from strings and orchestral instruments was very open, non-fatiguing.  And as the various movie clips progressed, I found myself really enjoying it.  So not the ultimate in natural timbre, but really nice for the enveloping surround sound.

Readers of this blog know I have a pair of (older) Spendor SP2/3 speakers, so to some extent I am a Spendor fan.  Well, I finally got to hear the Spendor D7s.  As my son said to me, “That’s the most natural sounding system I’ve heard this morning.”  And I agreed.  Am I going to tell you what speaker cables and interconnects were used?  No.  These shows aren’t the place to have a good critical listen, but you can get ideas for things you might want to hear more of at a later time.

What else?  I heard the TAD Evolution 1.  These are a sort of trickle down product from the very high end TAD Reference speakers.  The Reference speakers have a coaxial tweeter/midrange made of beryllium, and sound fantastic – heard them in Munich in 2012.  The Evolution has a similar coaxial tweeter/midrange, but the midrange cone is magnesium, a little less exotic.  I had just a brief listen, but I thought they sounded very good.

On a somewhat tangential note, Troels Gravesen recently did a study of the Seas C18EN001-M coaxial midrange tweeter, see: http://www.troelsgravesen.dk/COAX-18.htm.  We’ll see what he comes up with, but hopefully he will soon publish a passive 3 way design with the Seas coax and one or two woofers, perhaps a poor man’s TAD compact reference (http://tad-labs.com/en/consumer/cr1/) or something like the TAD Evolution 1.  Depending on what design Troels produces, that might be a future speaker for me to build.

More good things: I heard (again) the Volti Audio Vittora, and thought they sounded quite good.

Quintessence Acoustics (designed partly by Mr. PBN) had a large speaker with impressive bass, I would like to hear them in a more appropriate room, I think the room was way too small.  Next door was PBN speakers with Mr. PBN himself, Peter B. Noerbaek.  Just in looking online, I previously had thought of PBN as a bunch of overgrown monster speakers.  I have no other familiarity with the brand.  They were demoing (I think) a relatively modest size speaker, the SP-3, and whatever it was, it sounded quite good.  I complimented Mr. PBN, and something about his response struck me as very genuine – he seemed really pleased that I liked them, not all full of himself like “Oh of course my speakers sound good.”  Anyway, for what it’s worth, a very positive impression for me.

KEF has its new line of reference speakers, also perhaps a poor-mans TAD compact reference, the new Kef Reference One is a largish 3 way stand mount with a coaxial tweeter/midrange.  I was hoping to like these, but at the show they sounded rather thin/anemic.  I’ll probably try to hear them some other time under better conditions.  In contrast, a next door room had a pair of (I think) R900 speakers, they were much fuller and richer sounding.

Wilson Audio had speakers in several rooms.  I dunno what it is, I have yet to like any of them. Maybe it’s the demo song.  Maybe it’s the room.  Maybe it’s the partnering equipment.  Maybe you have to sit in just the right spot for them.  At any rate, given their fame and status among high end speakers, I expected to like them more.

I finally got a chance to hear a pair of Rockport Atria speakers, and I thought they sounded excellent.  And I had a quick listen to the Vivid B1 speakers, also excellent sounding.  See, not everything sounds bad to me!

A few things that are somewhat “different.”  First was a line of speakers from Sweden, Larsen Hi Fi named after John Larsen, who I presume was at the show (someone representing the brand was talking to visitors in the room).  This is creative – a line of speakers that are intended to work at their best right up against a wall, using the sound reflections from the wall to sound natural.  Apparently, Mr. Larson used to work with Stig Carlsson who influenced some past notables including Roy Allison and Peter Snell in ways to design speakers that work well near walls and other boundaries.  In my brief listen, I thought the speakers sounded quite good and were modestly priced by high end standards (somewhere around $2K or so).

Now it’s easy to feel like there are a glut of speaker companies out there, so when there is new speaker company you have to ask why bother? And what does this company do that is different?  It’s easy enough to make a somewhat decent sounding speaker – you can buy drivers (woofer, midranges, tweeters) from various companies, any cabinet maker can make cabinets, and sophisticated design software and measurement tools are widely available and relatively inexpensive.  I have commented before on the design work of Troels Gravesen and his diy speaker site.  While his knowledge and skill were acquired over many years, and he is undoubtedly a brilliant, creative, and extremely talented man, his designs are built in his home workshop and with modest measuring/test equipment.  One thing he does though is he sweats the details – it seems he can crank out a design quickly, but then he listens – really listens – to lots of music and tries out lots of things and refines and repeats.  How do I know this?  I know since I built two of his designs, and have thoroughly read his website and descriptions of his process and experiments.  But back to the point, it doesn’t take much to start a speaker company.  So when I encountered new speaker company Ryan Acoustics, I was curious – who are they and why another speaker company?

I had a short conversation with one of the founders, the company is two brothers who had a speaker company back in the 1980s, Ryan Acoustics.  Well, these guys must know something – they have a new line of modestly priced speakers that sound quite good.  Their speakers are hand crafted in America, and use their own woofers, midranges, and tweeters.  Now that’s rare – a start up speaker company that makes their own drivers.  Yes, one expects KEF makes their own drivers, and so does TAD and Rockport and Vivid, and more recently Sonus Faber, and of course the famous BBC legacy companies like Spendor and Harbeth make their woofers, but two brothers in California starting (or restarting) a speaker company to make their own drivers, that I didn’t expect. Well, they don’t exactly have a factory that makes the drivers, they have the drivers custom made to their design and specification, but clearly these are not minor mods to stock Scan Speak drivers (Scan Speak are excellent drivers, that’s not my point), but are really unique proprietary designs.

Ryan has 3 models: a little two way R610, a 2 1/2 way R620 (2 woofers and 1 tweeter), and a 3 way R630.  I heard the R620.  Now this is not some mega buck speaker, I think the price is in the $2K range, and I thought it was natural sounding with impressive bass.  In fact, the bass was beyond what I expected from such a slender and modest size speaker.  I wish them luck, it’s nice to see a company that wants to manufacture audio products in the USA, and no less a speaker company that actually designs its own drivers.

Finally, I heard the Polymer MKS-X speakers.  I didn’t expect to like them – their ads are all about how “high tech” the speakers are with diamond tweeters and diamond midrange.  I believe that excellent sounding speakers don’t necessarily require “exotic” materials.  And I thought it kind of cheesy that they tried to attract attention to their room with two tall/slender attractive women in slinky dresses – kind of like at the auto shows.  But this may have been the best sound I heard at the show, what can I say, the speakers sounded very natural and lifelike.

That’s it for now, I have some photos to add later.

 

 

Vinyl as a High Resolution Format

<Please note below that words in bold are links to other webpages>

I continue to be perplexed when vinyl is proclaimed to be “high resolution” with the implication that digital, and especially “redbook” digital, is not. For example, the Audiophiliac recently posted this. Now I took a previous shot at this discussion here. And I intend to write a more technical post later sharing some interpolation experiments I have been doing.

For now let me offer a few brief points:

1) Most vinyl today is produced from a digital recording. Arguments that claim vinyl (analog) audio is inherently superior to digital are clearly nonsense for cases where the final product (say a vinyl lp) has a digital audio master tape as its source.

2) Digital audio technology is often not well understood and incorrect arguments are made about its inadequacies. For example, I have read assertions that “vinyl tracks the analog signal exactly, while digital is quantized into steps.” Both analog and digital systems have noise, and digital quantization, handled properly, is another form of noise that is likely to be much lower in level than noise from an analog source. A lot of good information is written up as vinyl myths.

3) I like vinyl, and I have heard vinyl systems that sound really good.  But it is clearly an imperfect medium – the vinyl suffers from wear and tear, it gets dirty, etc.  In terms of dynamic range, I’ve read that vinyl has about 80db, clearly a lot less than 16 bit digital. When making a vinyl master from say a 24 bit digital recording a lot of processing has to take place to limit the highs/lows/softs/louds of the audio signal so that during playback the turntable’s stylus will not jump out of the groove; there are limits on what grooves can be successfully tracked. There is a well known equalization process (which surely must introduce noise and distortion) to reduce the bass and high frequencies for vinyl (see stereophile), and the overall sound is compressed – soft passages are made louder, loud passages made softer (more information).

4) In this blog I have made frequent mention of my high regard for Spendor loudspeakers.  Another very fine speaker company with a BBC pedigree is Harbeth UK, and I would wager a fair number of Harbeth owners are also vinyl owners.  There is a very sophisticated and enlightening discussion thread in the Harbeth User Group forum.

5) It is hard to objectively compare say vinyl and digital because often the recordings are different, or mastered differently, and one can put together a system that better flatters one medium or another.  A digital system is going to be more neutral in frequency response, so a system (cartridge, preamps, etc.) that flatters vinyl by deviating from neutrality in a pleasing way may sound bad with a more neutral system (e.g., digital).  I have digital and vinyl versions of certain recordings and in some cases I like the vinyl version better – not because there is something wrong with digital technology, but because I don’t like what the mastering engineer did to change the recording for digital or I do like what was done for vinyl.

A final comment, for what it’s worth.  About a year ago, I went to a boutique high end dealer, very nice fellow, I wanted to hear a certain brand of loudspeakers.  This shop was clearly “pro vinyl” and the dealer, in a non-pushy way, was trying to sell customers on his customized turntables and related equipment, and made frequent comments about the “organic sound” and “warmth” of vinyl. I brought in and played several of my vinyl records, and they sounded great, it was clear that his turntable/cartridge/phono pre-amp were a cut above my home system (albeit a lot more expensive too!). The dealer also had a vacuum tube DAC, and my compact discs sounded worse through his system than through my home system. If I had measuring equipment we could probably explain what was going on, it seemed to me though that the dealer had put together a system that was optimized for vinyl.

 

iPad (mini) Networked without Wifi!

In the previous post, I described how to use the iPad (or iPad mini) as a networked audio player. A key ingredient was to have a good wifi connection for the iPad. It never occurred to me that you can have a hard-wired (ethernet cable) internet connection for the iPad, but guess what, you can! I came across this article a few hours ago: http://9to5mac.com/2014/01/10/video-connect-your-ipad-to-the-internet-via-ethernet-cable-with-this-easy-hack/.

Now, the 9to5mac article describes connecting your iPad directly (using a powered usb hub and appropriate cables) to a router. I found that I could connect the iPad to a powerline ethernet adaptor (instead of directly to the router).

It looks like this:

Notice that there are 3 connections to the powered USB hub: the ethernet connection (via ethernet to USB adapter cable), the connection to the USB DAC (or in my case, DAC bridge), and the connection to the iPad (with USB cable plus USB to lightning adaptor).

Let’s review the components I used and connections:

*** Components ***
iPad (mini) with lightning port
apple lightning to USB adapter cable + USB cable
powered USB hub (I use D-link DUB-H4)
apple USB to ethernet adaptor cable + ethernet cable
Bel Canto mlink plus USB cable (to connect to powered hub) plus coax cable (to connect to DAC)
pair of powerline Ethernet adapters (one connected to router, one connected near iPad)

*** Connections ***
the powered hub is connected to an AC (power) outlet
the iPad connects to the powered hub via lightning to USB adapter plus USB cable
the powerline Ethernet near the router is plugged into an AC outlet, and then connected to the router with an ethernet cable
the powerline Ethernet near the iPad is plugged into an AC outlet, and then connected to the USB powered hub using an ethernet cable plus USB to ethernet adaptor
the Bel Canto mLink connects to the powered hub with a USB cable, and connects to a non-USB DAC with coax cable

The advantage of this approach over the wifi approach (previous post) is there is no need to have an Airport Express (or other wifi router) running bridge mode near the iPad. However, it can be an advantage to have Airport Express as it creates a local wifi that can be used by other devices as well as the iPad.

An iPad Mini as a Network Audio Player

I have it on my list of things to do  to go to Scotland and visit Linn Audio someday.  I’m not Scottish, but my grandmother was born in Glasgow, and I think the Linn company is innovative and one of the more interesting audio companies.  And they make their own stuff in Scotland I believe, so that’s cool.

If you visit the Linn site, they have a variety of “network music players” that (if I understand correctly) play digital music from a variety of sources, primarily with a NAS storage server for the music files and with a physical ethernet connection from the NAS server to the Linn player.  So, commonly, a NAS storage device  would be plugged into say the main router, and then ethernet cabling or powerline adapters are used to connect with the Linn network player.

Now I’m not arguing against purchasing the Linn players – they include their own DACs and are presumably high quality, but here’s a cheaper approach if you already have a DAC that gave good results for me.  I have my digital music files on an Apple Mac Mini, but it could be any Apple or Windows machine that can run iTunes.  Let’s call this iTunes library “Michael’s library.”  I configure the iTunes to use “Home Sharing.” Now it’s important to connect the host machine to your home network with a good connection, ideally plugging directly into the main router, or using a good wifi or powerline ethernet connection.

The rest of the solution is shown below:

 

A very important element of the solution is to have a good wifi signal for the iPad mini.  Until recently, I had lousy wifi in our main listening room, and one way to improve the wifi signal is described in the previous post.

The iPad mini also uses home sharing: if you go to the music app, the last selection on the bottom says “More”, click it and then you see a little house with a musical note inside with text caption “Shared.”  Click on Shared and then select “Michael’s Library” (in my case, your selection will be different).

So now the iPad has fast and reliable access to all the files in the main library, no copying or synching needed, and no need for extra storage or memory for the iPad.

So how do you get good sound out of the iPad?  

The most publicized way to get music from an iPad is to use Apple’s Airplay with say an Airport Express, and then the Airport Express can output a digital (optical) signal to a DAC. I’ve done that approach, and I really think it just doesn’t sound as good. The best explanation I have for the limitations of using Airplay are: 1) if you have high-res audio, using Airplay means the data is resampled and resized to lower resolution, and 2) the Airport Express becomes responsible for the timing accuracy of the digital data stream sent to the DAC, and I believe this makes a difference.

However, if you get the data from the iPad using a USB connection (camera connection kit), then you are not using Airplay, and the bits coming from the iPad are going to be the same bits as from any other device, the issue though is: a) are there errors in the transmission of the data from say the server to the iPad? (that’s why you need a good signal), and b) are the bits sent on their merry way from the iPad to the downstream DAC with the proper timing?  Good news, some clever people have made the timing of data from the iPad (or a computer) a non-issue using USB asynchronous mode transmission.  You can either feed the output of the iPad mini to a USB dac that operates in asynchronous mode (meaning the USB dac has its own clock, so the iPad’s timing becomes irrelevant), or, what I did, you use a USB to SPDIF converter, again, one that operates in asynchronous mode.

None of this information is secret, but it’s not well publicized either that:

a) An iPad can act as a server that “pulls” the music from any computer running iTunes

b) The music in iTunes accessed by the iPad can include lossless, and high resolution files (I tried both 44.1 KHz 16 bit and 88.2 KHz 24 bit loss ALAC files from Linn Audio, they sound great!)

c) The iPad can work with at least some USB DACs or USB “bridges”, but you likely need a powered USB hub. I purchased a D-Link USB powered hub from Amazon for less than $25, works great.

d) Because you may already have a good DAC in your audio system, and it may not be USB or compatible with the iPad (my situation), you can use a USB to SPDIF bridge which gets the data from the iPad with a USB cable, and converts it for use with a coaxial input (SPDIF) DAC.  I bought a Bel Canto USB converter (the mLink) and it works perfectly with the iPad (again, not publicized).

Let me know if there are questions or comments.  I believe an iPad used in this way has the potential to sound as good as anything else.  The data transmitted by the iPad, in theory (I don’t have lab equipment) should be bit perfect, and the timing accuracy of the bit stream is a function of the DAC (or combined bridge/DAC) and not the iPad.

I also tried the above setup with a laptop computer running iTunes, and achieved what I think are identical results.

Strategies for Better Wifi

This is an audio blog, so why digress into wifi?  Because it (wifi) is a convenient way to get access to your music.  The first challenge is to get a good strong signal to your media device (laptop, iPad, other tablet, etc.).

I recently came up with a strategy for better wifi in my house.  It’s not a patentable discovery, I’m sure other people have thought of it, but it wasn’t obvious to me and I’ll bet not too many people have implemented the strategy.

If you want wifi in your house, the “obvious” strategy is to get a wifi router, for example, Apple’s Airport Extreme or Airport Express.  I like Apple products, not to the point of my qualifying as a “fan-boy,” but I’ve had good success especially with their Airport products.  OK, so for the sake of argument, let’s say you have an Airport router, you set up your wifi network, but you find the reception is not so good in certain parts of your home.  What do you do?

You could get a newer or more powerful wireless router in the hope that it produces a stronger wifi signal.  The next possibility is to set up a “repeater router” that “extends” your wifi network.  For example, if you have an Airport Extreme, you can also get an Airport Express to “extend” the network.  I’ve had success doing that in some parts of my house, but not everywhere.

The strategy I stumbled upon is to combine power line Ethernet with a wifi router.

To go into more detail: the solution above requires a pair of powerline ethernet adaptors, available from manufacturers including Netgear and Linksys. I bought a pair of Netgear adaptors from the local Frys for a modest price of around $50.  The powerline adaptors use the electrical wiring in a house to transmit the ethernet data, with one powerline adaptor connected to the main router, and the other powerline adaptor plugged into a remote location in the house.  Then, you connect a wifi router, e.g., Apple’s Airport Express, into the second powerline adaptor.

The surprising thing to me is that you can actually have multiple wireless networks that inter-communicate, without “extending” each other.  So, as a variation of the above, let’s suppose that the main router in the home is also a wifi router.

 

There are a couple of tricks to getting the above configuration to work.  I will give details for configuring the Airport Express.  The same general ideas apply to other wifi routers.  The most transparent setup is to have the second wifi router (Airport Express) created with the exact same network name, security type, and password as the other wifi router(s).  Using Apple’s Airport Utility, you go into “Network”, then set “Wi-Fi Mode” to “Create a wireless network.”  Then you set the “Security”, and “Password” to match your other wifi network.  The next thing is to go into the “Advanced” menu, then the “DHCP and NAT” sub-menu, and set the “Router Mode” to “Off (Bridge mode)”.  That’s it.

So, in summary:

1) It is not well publicized that one use multiple wireless routers (in say bridge mode), connected to the same physical ethernet network, and create multiple wireless networks with the same name.

2) By using the wireless routers together with powerline ethernet adaptors, it is easy to create a good wifi signal anywhere in the house.  Said another way, if you have a part of the home that does not receive a good wifi connection from the main wireless network, you can place a powerline ethernet adaptor in that poor reception spot, and then create a new identically named wireless network by connecting a wireless router (e.g., Apple’s Airport Express) in bridge mode to the powerline ethernet adaptor.

In a subsequent post, I’ll explain how I used this strategy in my home to create a good sounding networked digital source.

 

Do You Hear What I Hear?

(mking the Seas Curv loudspeaker)

WordPress gives me a bit of information: what blog pages are viewed, what search engine terms were used to get to the blog, and what countries people are from.

Not a big surprise, most visitors are searching for something related to Spendor or Troels Gravesen, and the posts that get the most views are my comparisons of Spendor SP2/3 to the CNO, and the Spendor S3/5 to the Seas Curv.

I thought it would be interesting to see if I could let you, the reader, hear what I hear. That is, I wondered if I could record my speakers and see if you can hear the differences I hear. The problem is that I don’t have a recording studio, and the only way I know to host a soundfile is to make a video slideshow with music on youtube.

So, here is my first experiment – a few minutes of music reproduced by the Seas Curv loudspeaker, and recorded with a Sure SM-57 microphone. I picked music that I thought would demonstrate a few things: naturalness of voice (both soprano range and lower male voices), some instrumental including brass, saxophone, piano, and drums. I used an E-MU 0204 to take the signal from the microphone, digitize it in 24 bit, and send it to my Apple laptop computer. I edited (cut) the sound samples with GarageBand, and maintained full resolution until the final export.

It is important to realize that the microphone itself has a frequency response curve that is not flat – the low notes are rolled off, and the high notes are accentuated. The middle range though appears to be very flat and accurate.

This is my first attempt, so I will get better at it. I plan to go back and repeat the recordings with the Spendor S3/5 so you can compare what is captured by the microphone and “hear what I hear.”

*** Update 9/11/2013 ***
Although I believe a short sound clip for educational purposes is “fair use”, someone or some system recognized the source of one of my clips and I had to remove for alleged copyright violation. Also, I have been having difficulties with the E-MU 0204, so I have not been able to make further recordings. Hope to try again soon.

Video: Not currently On youtube

California Audio Show 4 (2013) – A Visit with my Camerman

This past Saturday, I sent my fierce Corgi out to hunt for the California Audio Show:

Ah – there it is – the lovely Westin Hotel:

This time, I went with my trusty camera man in tow, so I have a few photos to share. My usual gripes about audio shows aside, I had a fun time and would recommend this show to locals.

OK, I just love the look of the Sonus Faber Olympica loudspeakers. This is a new line of speakers that made their debut a few months ago in Munich, and here is the Olympica 1 stand mount speaker, what a beauty!

This is just a gorgeous gorgeous speaker, with solid walnut wood for the top and bottom, stitched leather on top, and leather covered baffle and rear, and a swoopy, curvy lyre shape that is a Sonus Faber signature. You can read more about them on the Sonus Faber site: http://www.sonusfaber.com/en-us/products/olympica-i.

I was told the speakers will have a retail price (in the USA) of $6,500, and another $1,200 for the matching aluminum stands. Pretty pricey in my book, but oh, they are lovely. Now value in high end audio is something of an iffy proposition, and given the high quality materials, limited production, and relatively costly Italian craftsmanship, others may perhaps view these as reasonable. For me, I feel confident that, if given the components (drivers, crossovers) of the SF speakers, I could make an externally plain box (but with interior complications and lots of bracing) that would sound quite similar, but it wouldn’t look this good! Seriously, it is hard for me to imagine that a fairly small 2-way loudspeaker with stands could be worth nearly $8K sonically – each speaker has a single 6″ paper (“cellulose”) cone woofer, and a more or less conventional tweeter. In my opinion, and based on my recent experience with DIY loudspeakers, it’s hard for me to justify such stratospheric prices when I can build something like a monitor version of Troels Gravesen’s Seas CNO for under $2K /pair even factoring in high end Excel drivers from Seas, paying a cabinet maker, and using top quality plywood, damping, and crossover components.

How did the Olympica I sound? Not sure, to be honest. They were being demoed with some techno music “noise” – I wonder what Franco Serblin (RIP) would have thought of such a choice of music for his brand. How can one judge the naturalness of speakers playing loud synthesized techno music? The one impression I had was that, on the one hand, the speakers made a lot of sound and bass for their size and relatively small woofers, but it wasn’t my idea of good bass, it was “impressive for a little speaker” bass, if that makes sense. I wish the demo had been better.

What else was disappointing? I had my first listen to the Wilson Audio Duette 2s, and they are, to me, an even worse offender when it comes to value, approaching nearly $20K with their matching stands. How did they sound in the demo room to me? Well they played loud, they did sound “big”, but the bass was boomy, and the sound was not natural. Another disappointment in my view was a pair of A Capella horn speakers with a fancy plasma tweeter. The sound was overly bright/harsh, and I can’t imagine that having a large midrange horn in front of a tweeter doesn’t cause all sorts of diffraction problems.

BTW, if you disagree with my opinions, well, that’s fine. If something does sound good at a show, I take that as a positive sign. If it doesn’t sound good, well, could be the room, could be the demo recording, could be the pairing of the equipment, and so on, I can only give my impressions, and I think it’s useful perhaps to read a different point of view to the professional sites where critical comments are less common.

Ok, so what did sound good at the Show? One of the best sounding rooms was from Music First Audio, they have a “transformer passive preamplifier” and were playing an analog reel-to-reel tape with Audio Note loudspeakers, the volume was reasonable, the recording was excellent, a very good sounding room! And I heard a demo of Vivid Audio Giya speakers – they sorta look like huge Hershey’s kisses, the sound was excellent, clearly a few notches above my home system in terms of dynamics and clarity at higher volumes.

One room that impressed me with good sound for a more down to earth price was a demo of AudioPhysic Classic 20 loudspeakers powered by Nait XS2 integrated amplifier. The AudioPhysic speakers are slender 3 way towers with side-mounted woofers, I don’t recall exact pricing but somewhere between $4-$5k, and the Nait XS2 is around $2,500, downright reasonable compared with some rooms where the price of the components exceeds the cost of many houses in the USA.

I want to mention a few turn-offs. The show was not very friendly for young folks. My teenage son desperately wanted to hear “his” music instead of classical or older pop/rock. What is his music? Nothing too unusual for today’s teenagers – he’s a bit eclectic, but as a random sampling of things he’s played for me, he’ll listen to Jay-Z, Justin Timberlake, Lady Gaga, Rhianna, and so on. In the majority of rooms, nothing remotely like his music was being played. Finally, we arrived at a demo room, the host asked what we wanted to hear, my son, seeing his playlist, said “Coldplay” and then the host  proceeded to play something else – hello – he asked for Coldplay! Folks in the audio industry, if you are reading this, the teenagers don’t have the money to buy now, but if you want them to become interested in audio, make the show more accessible to them!

The best part of the show for him – the big room with all the headphones and headphone amplifiers. We visited our friends at Woo Audio, they again had the lovely fireflies amplifier:

And here’s the trusty camerman enjoying the sound of Sennheiser HD800 headphones:

One more cool thing – I saw and heard Mr. Speakers headphones featuring a headcup that was made by a 3-D printer.

I prefer speakers to headphones, but no question, the biggest bang for the buck for good sound are headphones – get something like the Sennheiser HD800 and the Woo Audio fireflies amplifier, pair with your laptop computer, great sound!

Thanks for reading…

The Puzzle of HHT versus HTH

Are you open minded?  A classroom experiment I witnessed some years ago suggests probably not.  In this blog I am discussing subjects that may not have widespread agreement, such as 3 speaker versus 2 speaker playback, or thoughts about analog and digital audio.  And I wonder whether people really read and think about what I write, or if they ignore new information or ideas that might disagree with their existing beliefs.

More than 20 years ago, I attended a class by Professor Ron Howard at Stanford on “Decision Analysis” – I just checked and he still has a web page at Stanford so I guess he’s still teaching: https://engineering.stanford.edu/profile/rhoward.

In one lecture, Professor Howard posed a problem which, if I recall correctly, was titled something like “HHT versus HTH.”  The problem Howard posed is interesting because, in the classroom setting, and without a few quiet moments to reflect and think it through, it is a difficult one for most students to solve, including the bright master’s degree level students at Stanford that were the bulk of the class.  Therefore, and this is what was so fascinating, what is actually a question with a definite right/wrong answer takes on the characteristics of a subjective issue where people can “agree to disagree.”

Here’s the problem: two students (A and B) are brought to the front of the class, and each one is given a medallion with 2 sides – one side is “H” (head) and the other “T” (tails), and assume that in tossing the medallion the probability of H or T is equal.

Student A is to keep tossing the medallion until the sequence H then H then
T is achieved. 

Student B is to keep tossing the medallion until the sequence H then T then H is achieved. 

Question: who do you expect to achieve their goal in the least number of tosses: Student A or Student B or is it equal?

Of course, the interesting thing here is NOT the question, but the classroom
activity.  What Prof. Howard did after posing the question was to poll the class – how
many think A? how many think B? how many equal? how many are undecided? The
votes were recorded.  Then he asked a student in the audience to explain why “A” and another student to explain why “B”, etc.  Then he polled the class again, and again recorded the votes.The above process of listening to student explanations, polling the class, and recording the votes was repeated several times.  During the process, a lot of absurd and wrong statements were made, but eventually, one student figured out the problem, and stated the correct answer with a clear, logical explanation.

You might think that after the student stated and explained the correct answer, all the votes would switch.  But that did not happen.  The interesting thing?  Nobody listened.  That’s right, the votes remained fairly stable – people made up their minds initially, and then paid basically no attention to anything that was said.

Listening to the lecture and classroom participation/response was a
fascinating experience for me. Oh, and for those who care, it’s student B
who would be expected to first achieve the goal sequence.  Quick explanation: the two sequences have different symmetry properties, so that if we suppose both A and B miss on the second toss, A (who tossed HT) needs the third toss to begin his sequence, but B (who tossed HH) can start his sequence with the second toss, so B needs fewer tosses on average to achieve his sequence.

Digital Audio Basics: Stair Steps and Sample Rates

I took a prior stab at discussing vinyl versus digital audio, and I want to do a better job, but it’s going to take a bit of background discussion first.  I feel that to have an understanding of digital audio, and to discuss comparisons between say Vinyl and CD, or CD and high resolution audio, the concepts of sample rate and sample bits in digital audio need to be explored first.

I’m both excited and embarrassed to be writing this post.  I’m not Joe on the street, many years ago I took classes in digital signal and image processing, I earned graduate degrees in engineering from a prestigious university (although it has been a few years….), and yet I realize I didn’t have an intuitive grasp of some basic concepts.

I find that for myself, and for many others, concepts around digital images are more intuitive than concepts around digital audio.  For digital images, most of us know that a digital image is composed of little squares or pixels, each pixel is a single color, and the more the pixels the better the picture.  The color accuracy of a pixel is determined by the number of binary digits (bits) – the more the bits, the finer the gradation of a color from light to dark. A digital audio file is a lot like a digital image file, except that a digital image file has color data for points in space (the pixels), and a digital audio file has signal data (corresponding to the sound level) for points in time.  With digital audio, we have sound samples (corresponding to pixels), and binary data (bits) that determine the gradation of sound from soft to loud.

Most of us expect that, in audio, we get closer to the original signal by having more samples and more bits.  There are two issues that make our intuition not fully correct.  The first issue is the limited range of human hearing, and the second issue is the presence of noise, both naturally in our environment, and in all recorded audio.

First I’m going to talk about sampling rate, or how many audio samples are collected say every second.  The second issue, how much data (bits) is stored with each simple, will be considered in a separate post.  Let’s consider what happens when we sample a pure tone, say a sine wave.  Here I am not concerned with the precision of the sample, so we are not thinking about measurement error, round off error, or “quantization” that happens from using a finite amount of data to capture each sample.  Many readers will likely have seen a graph like below; suppose the smooth red curve is the original signal, the blue “stepped” curve shows the digitized version with say 10 samples, and the light green “stepped” curve shows the digitized version with say 20 samples.

A graph like the one above is a source of a lot of wrong thinking about digital signals.  What the above implies is that 10 samples gives a crude approximation of the original signal, 20 samples gives a better but still crude approximation, and that if we continued to increase the number of samples we would eventually get to a pretty good approximation of the original signal.  But this is completely wrong!

There is a famous theorem, the Nyquist-Shannon sampling theorem, and I recently realized (courtesy of a video from Xiph.org), that I didn’t truly grasp it.  What the Nyquist-Shannon sample theorem says is that if the frequency content of the original signal is limited to some maximum, say M (e.g., say 20KHz for human hearing), and if we collect samples at frequency 2M, then we can perfectly reconstruct the original signal.  In the example above, then, going from 10 samples to 20 samples does nothing to improve the accuracy of the reconstructed signal, as the original signal was a pure sine wave  and with 10 samples we are already way over the minimum of twice the frequency of the original signal.

So why does it go against intuition to think that more digital samples means a closer approximation of the original signal?  Probably because (most of us) we tend to mentally think of a stair step reconstruction of the original signal from the samples! Where do the stair steps come from?  From the idea of sample and hold.  Remember, a sample is the signal value at a particular time, but we don’t know (or record) the signal value in between samples.  So the simple thing to do is just imagine that the signal does not change values between samples, which is how I drew the above graph.

The reconstruction of an analog audio signal from digital samples though, does not use sample and hold for the final result.  The conversion process is more like a kind of curve fitting, requiring there to be a smooth output that goes through the sampled values.  If there were sudden stair step jumps in the reproduced signal, those sudden jumps would have high frequency content beyond our maximum frequency M.

To demonstrate, and this is not any sort of rigorous proof, but a way to improve intuition, I used Microsoft Excel, and I created 3 data sets for a sine wave: the first data set (red) represents the original signal (actually with 100 points), then 20 sample points, then 10 sample points.  I displaced the curves vertically so it is easy to see, they all have the exact same shape, and that is what happens when a digital audio file is reconstructed to be an analog signal – one gets a smooth signal with no stair steps, and increasing the number of samples does not improve the smoothness of the final signal:

Again, for most of us, this result is highly counter intuitive – surely more samples must make for a smoother curve.  The point though is that we have taken samples of a bandwidth limited signal – a signal that has maximum frequency content M.  If there was something “unsmooth” or highly irregular looking going on between the samples, that would imply higher frequency content.  Since we are sampling at 2M or greater, we have captured enough information even with the “crude” 10 samples.

Said another way, if I had not displaced the three curves vertically, they would lie exactly atop each other, and THAT is what the Nyquist-Shannon sampling theorem states: we can exactly reconstruct the original signal from the sampled data provided we have sufficient samples, and sufficient for this example would be only 2 samples!

A brilliant demonstration of the above is given by Monty Montgomery in the following video:

http://xiph.org/video/vid2.shtml

or on youtube at

Now one question in my mind, and perhaps yours, is what about measurement error and measurement precision?  I recall taking a basic lab course many years ago, and one of the first things drilled in to me (not talking about digital audio here) is that all measurements have limited accuracy.  Suppose I want to use a ruler and measure the length of a piece of paper.  The ruler has tick marks say every millimeter, or every 1/16 of an inch, so that limits the precision of my measurements.  Not only does the ruler have limited accuracy, if I make multiple measurements, or if more than one person makes measurements, it is unlikely that we get the same value every time.

In a future post, I’ll touch on the issue of measurement error and use Microsoft Excel to create a simple spreadsheet and graph the effect of timing errors (recording a sample say at .1003 seconds instead of exactly at .1 seconds), and the effect of using varying measurement precision (more or less bits) for the samples.

Thanks for reading!