One solution that immediate comes to mind is that the major software vendors should be encouraged to 
certify hardware that is proven to work with their software.

George, do you think the problems you documented come about from editing program - hardware driver 
interaction or is it more often than not purely editing software simply mangling data?

And this also raises the question, for the small/budget-starved institution, maybe it IS a better 
idea for them to record directly to CD media, on the assumption that a dedicated CD recorder used 
with good-grade media is less likely to mangle its own data?

-- Tom Fine

----- Original Message ----- 
From: "George Blood" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Monday, August 18, 2008 9:38 PM
Subject: [ARSCLIST] The Hope of Audacity...

> Hi, Friends,
> There are a variety of issues that have been raised in this thread.
> First is the disappointment others have mentioned that the tools we're  buying are not doing what 
> we think they're doing.  Namely, we're  feeding them data (output of the A to D converter) and 
> that data is  being changed between the input and being written out into a file.   Given the 
> commercial pressures we all understand, writing stable code  that has a few "quirks no one will 
> find" could be seen as a simpler  solution to making payroll than digging deep into the myriad of 
> problems and the cooperation necessary from other vendors.   Interoperability is a non-trivial 
> issue.
> Second, this is more than just an issue of you get what you pay for.   Sure Sonic and Pyramix tend 
> to perform better than the desktop  applications.  But they control much more of the signal chain. 
> The  lower priced programs hit their price point by handing off (or  depending upon or using) the 
> work of others (WaveLab, Samplitude,  SoundForge, etc. let M-Audio, RME, PreSonus, etc build 
> interfaces, for  instance).  But who wants to be working 20 different other companies  to assure 
> interchangeability.  Ideally, everybody does.  The reality  is, given the number of hours in the 
> day, are you going to add the  cool new feature, or chase down a bug with somebody else's code. 
> Apple tried to make the world a simpler place with CoreAudio.  Great  idea.  When it works.
> Yes, they should begin with integrity of signal.  But if that's an  obstacle, they build a 
> workaround (a "trap").  And it suits the needs  of 99% of their users (and appears to have 
> snookered a far larger  percentage!).  Who are we to deprive a bezillion garage bands of the 
> features they need and pay for, just so we can have the needs of our  small numbers met?
> Lacking the tool vendors doing this work, each of us needs to shake  down their own system.  It's 
> a royal pain in the butt.  Every change  in the system means re-checking it.  OK, so Peak 6 will 
> now preserve  the bext chunk (it used to blow it away).  But now it embeds waveform  display info 
> in the INFO chunk.  That's perfectly acceptable according  to the .wav spec.  But it means there's 
> garbage there that, 25 years  from now, some preservation is going to sweat recovering, because 
> "if  it's in the file it must have been important".  Every program, every  OS, every hardware, 
> every driver change has the potential to disrupt a  perfectly crystalline signal chain.
> Third, what does it tell us in the trade that, for all our efforts to  create higher resolution 
> files, there's been overwhelming satisfaction  with the lesser results.  Well, for one thing, it 
> reminds us that  there's more impact from good analog playback than from spending an  extra 
> whatever on 8 extra bits that are 30dB below the noise floor.   For another, good AtoD design does 
> amazing things when outputting  lower res files.  It reminds us that so many of our source 
> materials  are of such low (or lower) quality that there isn't more information  to be captured at 
> the higher resolutions.  Once we argued over 48k vs  44.1.  Then 96k.  Some want 192k, and others 
> think 384 is worth  doing.  Look folks, if you keep increasing the sample frequency, you  won't 
> start capturing video off the audio cassettes.  But you are  building more and more fragile 
> systems.  Where you have to break the  files into multiple sub-file (due to the 2GB file size 
> limit of WAV),  you create a chance to lose one of the pieces.  Is this really better 
> preservation than having one file (even 44/16) of the whole thing?
> Fourth, while the pursuit of ever higher standards is a worthy goal  (and the reason we have 
> converters that perform so well at lower  sample rates), there will always be circumstances in 
> which an  institution simply will not have the resources to "do it right".  We  can all salivate 
> at the enormous volumes of audio out there to be  saved.  As much as we dream of driving Lexuses 
> into retirement, your  garden variety historical society with a shoebox full of oral  histories 
> may never be able to justify spending "what it costs" to  "properly" digitize them.  If their 
> board can get its act together and  apply for grants, maybe there's a better use of that few 
> thousand  dollars. A use closer to their primary mission.  So a nice retired  couple volunteers to 
> bring in the CD-R recorder they bought for their  granddaughter (who really wanted an iPod), plugs 
> in the cassette  player they picked up at a garage sale, and digitizes the shoebox of  cassettes. 
> And lo and behold, that bit of history is preserved for  the future.  Did they only get 80% of the 
> information off the tapes.   Yeah.  But 80% is a lot more than nothing.
> This is a very slipper slope, of course.  For every Harvard or IU (and  others) here are plenty of 
> large institutions buying crap equipment,  giving it to sleep-deprived, hung-over work-study 
> students.   Institutions that really do know better, and who could manage the  resources to do a 
> much better job (even if not "do it right").
> And so it goes.  So it always has.  And so it always will.  Our job is  to help advance the trade, 
> do as well as we can, at a price that means  more of it gets done.  Until we're all using $25,000 
> workstations and  $15,000 A to D converter, (and more hours making our systems pass 24  bits than 
> lingering on ListServs!) we shouldn't be pointing fingers at  the poor blokes trying eek out a 
> living selling $250 (or free)  software.  You get what you pay for.  If you paid for that, you got 
> it.  Of your own free will.
> Back to your regularly scheduled madness.
> G
> PS I bought my Prius when gas was $2.00/gallon.
> On Aug 18, 2008, at 12:00 AM, ARSCLIST automatic digest system wrote:
>> I believe George Blood is on this list.