The $10000 Hi-Fi

On 5/5/2015 1:54 PM, Tim Wescott wrote:
> On Mon, 04 May 2015 17:38:30 -0400, rickman wrote:
>
>> On 5/4/2015 5:23 PM, Greg Berchin wrote:
>
> < snip >
>
>>> Don't ask me for a definition of "organic", either. I don't have one.
>>
>> You and no one else.  It's a "Through the Looking Glass" term meaning
>> exactly what the speaker intends it to mean, nothing more and nothing
>> less such that in the end it says nothing at all.
>
> My slightly less cynical take on this is that the term means "I'm doing
> something loosey-goosey and I don't know what, but I can't deny that it
> works".
>
> So it says _something_, just not anything specific enough to satisfy an
> engineer.

 From what you wrote I'd say it still means nothing... "I don't know 
what"???

-- 

Rick

spope33@speedymail.org (Steve Pope) writes:

> Cedron <103185@dsprelated> wrote:
>
>> Back in the day, the most important component to sink your
>> money into were the speakers.  I don't know if that is the case
>> any more, I suspect it is still true.  So, if you are comparing
>> DACs, you have to have responsive enough speakers.
>
> Depends on the DAC.  Some of them are so bad (i.e. DAC's in the cheaper
> sound chips found in certain phones, CD/DVD drives, laptops, DVD players) 
> are so bad that the low quality is audible almost regardless of the rest
> of the signal chain.

Care to name a chip or two?

I've never been aware of a device in the past 20 years that sounds
anywhere near bad. No, my hearing is not what it used to be...
-- 
Randy Yates
Digital Signal Labs
http://www.digitalsignallabs.com

[...snip...]
>
>I do recordings of amateur music, such as high school orchestra
>concerts in 24 bit WAV, then convert to 16 bits. That allows me to
>turn the record level down a little, to make room for surprising peaks.
>(Timpani or cymbals, for example.)
>
>I then select which 16 bits to use, after adding appropriate dither.
>
>Most likely the background level is high enough that it doesn't matter,
>but it isn't hard to do 24 bit at 44.1kHz, then convert to 16 bit CDs.
>Conveting 48kHz to CD is a lot more work.
>
>-- glen

Is the dither significant soundwise?  In other words, can you hear the
difference with or without?  It would seem to me if you were merely
shifting a certain number of bits, or dividing my some factor, the lowest
order bit wouldn't really matter.

I am currently rewriting my Windows based "taper program" in Linux.  My
plans are to stick with 44100, 16bit, but your approach is causing me to
reconsider.

Do you apply software audio compression? (not data compression).  Many
years ago I derived the formulas for a compression curve that extended the
straight line portion with a hyperbola that matched the first derivative
so there was no knee effect.  I am considering building it into my
recording program, but not sure it would be worthwhile or better to leave
it as an external process.

I tend to record rock bands in bars, so the dynamics aren't as great as
symphonies.  They are generally full on loud.

As for earbuds, I don't like them.  It is too easy to damage your hearing
and you just can't feel the bass in your gut.

Ced


---------------------------------------
Posted through http://www.DSPRelated.com

On Tue, 05 May 2015 15:21:43 -0400, rickman wrote:

> On 5/5/2015 1:54 PM, Tim Wescott wrote:
>> On Mon, 04 May 2015 17:38:30 -0400, rickman wrote:
>>
>>> On 5/4/2015 5:23 PM, Greg Berchin wrote:
>>
>> < snip >
>>
>>>> Don't ask me for a definition of "organic", either. I don't have one.
>>>
>>> You and no one else.  It's a "Through the Looking Glass" term meaning
>>> exactly what the speaker intends it to mean, nothing more and nothing
>>> less such that in the end it says nothing at all.
>>
>> My slightly less cynical take on this is that the term means "I'm doing
>> something loosey-goosey and I don't know what, but I can't deny that it
>> works".
>>
>> So it says _something_, just not anything specific enough to satisfy an
>> engineer.
> 
>  From what you wrote I'd say it still means nothing... "I don't know
> what"???

Since I'm not in the habit of dismissing the phrase "I know that it 
works", even if it's followed by the phrase "but I don't know why", I 
think the term has value, and contains a considerable weight of practical 
meaning, even if it's not pleasing to the engineering personality.

-- 

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Randy Yates  <yates@digitalsignallabs.com> wrote:

>spope33@speedymail.org (Steve Pope) writes:

>> Cedron <103185@dsprelated> wrote:

>>> Back in the day, the most important component to sink your
>>> money into were the speakers.  I don't know if that is the case
>>> any more, I suspect it is still true.  So, if you are comparing
>>> DACs, you have to have responsive enough speakers.

>> Depends on the DAC.  Some of them are so bad (i.e. DAC's in the cheaper
>> sound chips found in certain phones, CD/DVD drives, laptops, DVD players) 
>> are so bad that the low quality is audible almost regardless of the rest
>> of the signal chain.

>Care to name a chip or two?

>I've never been aware of a device in the past 20 years that sounds
>anywhere near bad. No, my hearing is not what it used to be...

"Sound chip" is perhaps a misnomer since the functionality is
now (usually) integrated into a much larger ACIC.

However -- my Nokia 5310 phone ("Xpressmusic Series") definitely
has better sound that a more recent Sony phone (Xperia series).
Nokia reportedly developed a sound chip especially for that series.

The difference among CD or CD/DVD drives with an 1/8" audio
output jack (of the form factor that one would install in a desktop 
computer) is also very noticeable.  I have had Sony and Toshiba
which sounded very good compared to others.

Of course I could be imagining the audio quality differences, but
I think not -- typically I take the device's 1/8" output and plug it 
directly into my system, which consists of a reputable vintage
Kenwood amplifier and Klipsch speakers.  So there's not much
to blame if the audio quality is off, other than the sound chip
(or equivalent function) in the device.  

Steve

Cedron <103185@dsprelated> wrote:

(snip, I wrote, regarding 24 bit WAV files)

>>I then select which 16 bits to use, after adding appropriate dither.

>>Most likely the background level is high enough that it doesn't matter,
>>but it isn't hard to do 24 bit at 44.1kHz, then convert to 16 bit CDs.
>>Conveting 48kHz to CD is a lot more work.
 
> Is the dither significant soundwise?  In other words, can you hear the
> difference with or without?  It would seem to me if you were merely
> shifting a certain number of bits, or dividing my some factor, the 
> lowest order bit wouldn't really matter.

I suspect no-one would notice, but when I first added it I did
some tests.  Since I can shift different amount, I could make the
resulting signal pretty small. Then I (carefully) turned up the
volume control until I could hear it. At very low (after shift)
levels, the noise is audibly better with dither.

But live recordings with kids in the audience tend to have enough
background noise, in addition to ventilation systems not designed
for low noise, that it probably doesn't matter for real signals.

I use a pretty simple LFSR generator, based on CRC32, to generate
bits, shift and add.
 
> I am currently rewriting my Windows based "taper program" in Linux.  My
> plans are to stick with 44100, 16bit, but your approach is causing me to
> reconsider.
 
> Do you apply software audio compression? (not data compression).  Many
> years ago I derived the formulas for a compression curve that extended the
> straight line portion with a hyperbola that matched the first derivative
> so there was no knee effect.  I am considering building it into my
> recording program, but not sure it would be worthwhile or better to leave
> it as an external process.

No. The only thing I do is select the shift amount for each track.
I compute the peak and RMS (over the whole track) values, then decide
on the shift amount from that.
 
> I tend to record rock bands in bars, so the dynamics aren't as great as
> symphonies.  They are generally full on loud.

In that case, you probably won't notice.
 
> As for earbuds, I don't like them.  It is too easy to damage your hearing
> and you just can't feel the bass in your gut.

-- glen

Cedron <103185@DSPRelated> wrote:

>Is the dither significant soundwise?  In other words, can you hear the
>difference with or without?  It would seem to me if you were merely
>shifting a certain number of bits, or dividing my some factor, the lowest
>order bit wouldn't really matter.

The difference between dithered and undithered quantizing is
definitely audible on constructed test signals (e.g. tones).
Undithered quantizing creates correlated noise, which on
a single tone signal is readily audible, even at 16 bits precision.
The "noise" becomes tone-like, resembling (a) beat frequency(ies).

Whether it would be audible on a real-world signal, I can't say.
But if there is no cost or low cost to add the dithering, then just
dither it since it is one less thing to worry about.


Steve

On 5/5/2015 4:31 PM, Tim Wescott wrote:
> On Tue, 05 May 2015 15:21:43 -0400, rickman wrote:
>
>> On 5/5/2015 1:54 PM, Tim Wescott wrote:
>>> On Mon, 04 May 2015 17:38:30 -0400, rickman wrote:
>>>
>>>> On 5/4/2015 5:23 PM, Greg Berchin wrote:
>>>
>>> < snip >
>>>
>>>>> Don't ask me for a definition of "organic", either. I don't have one.
>>>>
>>>> You and no one else.  It's a "Through the Looking Glass" term meaning
>>>> exactly what the speaker intends it to mean, nothing more and nothing
>>>> less such that in the end it says nothing at all.
>>>
>>> My slightly less cynical take on this is that the term means "I'm doing
>>> something loosey-goosey and I don't know what, but I can't deny that it
>>> works".
>>>
>>> So it says _something_, just not anything specific enough to satisfy an
>>> engineer.
>>
>>   From what you wrote I'd say it still means nothing... "I don't know
>> what"???
>
> Since I'm not in the habit of dismissing the phrase "I know that it
> works", even if it's followed by the phrase "but I don't know why", I
> think the term has value, and contains a considerable weight of practical
> meaning, even if it's not pleasing to the engineering personality.

Ok, I'm not sure what it has to do with the "engineering personality". 
This is simply a question of meaning and facts.  When someone says, "it 
works", that means in their often very limited experience.  I don't 
think I've ever heard the word organic used to mean anything even close 
to how you use it.  So if that is what it means to you we have yet 
another definition to add to the mix.

As a counterpoint to your "engineering personality" reference, my 
roommate is just as interested in eliminating the BS to separate the 
truth from the superstition, yet he was a Political Science major and 
has never had a technical job.  He often makes fun of the sort of people 
who use terms like  "organic".

-- 

Rick

On Tuesday, May 5, 2015 at 2:21:44 PM UTC-5, rickman wrote:

>  From what you wrote I'd say it still means nothing... "I don't know 
> what"???

Well, the bottom line is that I was trying to make a joke. Some things are not easily quantified, or even explained. They just come with experience. If you really want a definition, I will try.

It is my belief (asserted without proof, as it is an opinion) that the best-sounding digital audio systems are those that emulate analog systems, where possible. For example, there are filters that are so arithmetically complex that, despite perfect implementation, they don't sound very good. Sometimes drawing the line between an ingenious implementation that would be impractical in analog, and just having "fun with numbers", is best accomplished by listening. The "organic" part of this refers to backing-off of that "fun with numbers" implementation and going with something not quite as "perfect" that actually sounds better.

Some examples?

In an old issue of Stereophile there is an article describing what a certain high-end manufacturer did to change one of their DAC products from an average-sounding product to a superior sounding product. I don't remember the exact context, but I do remember the change: the switched to rounding at the 24th bit to truncating at the 24th bit. Conventional wisdom said that the difference at the 24th bit should be essentially inaudible ... or that, if audible, the rounding should sound better than the truncation, but the opposite was true.

I designed a set of digital inverse-RIAA filters for a high-end product. Instead of applying a bilinear transform to the s-domain transfer function, which is how most such filters are designed, I designed a set of filters that modeled both magnitude and phase. In terms of traditional performance metrics, the magnitude response of my filters might not have been quite as good as the bilinear transform. My phase response was better, but "everybody knows" that phase response is inaudible, right? Well, my filters earned reviewer comments like, "I did not believe a digital phono stage could sound so natural and it constantly brought out new detail on very familiar material. To my ears it combined the clarity of very good high res digital with the ease and richness of vinyl. [...] the double bass was tangible and you could hear the resonance of the wood as if in a club. I got massive amounts of previously unheard detail and a deep soundstage [...] turned into an amazingly realistic experience. You could hear each musician in a symphony orchestra playing together as individuals each contributing to the overall sound. On a half way good system you can distinguish the different brass instruments and the woodwinds when they play in unison, but this was the first time I have ever heard and been able to separate out the individual violinists playing in unison. It was the closest I have ever heard to a live orchestra."

That is the closest I can come to a definition for "organic" in this context.

[...snip...]

>I don't remember
>the exact context, but I do remember the change: the switched to rounding
at
>the 24th bit to truncating at the 24th bit. Conventional wisdom said
that
>the difference at the 24th bit should be essentially inaudible ... or
that,
>if audible, the rounding should sound better than the truncation, but
the
>opposite was true.

This is the crux of what I was really asking.  If diddling with the 24th
bit is audible, then certainly diddling with the 16th bit would also be
audible (as Steve Pope asserted).  I still find it hard to believe.

[...snip...]

>
>That is the closest I can come to a definition for "organic" in this
>context.

Perhaps, "authentic" would have been a better word choice and saved you
from being nitpicked.  Just guessing.

Ced


---------------------------------------
Posted through http://www.DSPRelated.com