![]() ![]() |
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
In Reply to: Re: Thanks for the clarification about your perspective.... posted by John Atkinson on October 01, 2003 at 10:51:15:
The Tascam is able to write both audio and computer data (audio mode and USB mode). Would this mean that the two data writing processes are different so that less care is used when writing in audio mode ?Furthermore, what about the claim that anything in the recording chain affects the sound (for those studios which use cables like Analysis Plus) ?
Are you saying that because of the lower quality in writing audio data subtle differences are likely to be lost ? What about greater differences like CD vs MD or different capacitances of the phono stage ? What about the Sony walkman vs Wadia scenario ?
This could indeed be a valid technical reason that speaks against such a test. However if someone manages to score 100%, would this prove something ?
Follow Ups:
If I were you Klaus I'd forget about trying to 'prove' anything conclusively.Tests in almost every area you could imagine tend to 'suggest' a certain conclusion, with statements such as 'high correlation' and 'statistically significant' being the order of the day; 'proof' is almost always elusive and hence tends to be subjective.
> The Tascam is able to write both audio and computer data (audio mode
> and USB mode). Would this mean that the two data writing processes
> are different so that less care is used when writing in audio mode ?
Yes. Which is why a data CD-R has less storage area than an audio
CD-R with the same nominal capacity. Take an 80-minute CD-R blank
and burn it as an 80 minutes audio disc using uncompressed WAV or
AIF files. Now take a second 80-minute blank and burn the same audio
files on it as computer files. You won't be able to fit them all
on the disc!
> Furthermore, what about the claim that anything in the recording
> chain affects the sound (for those studios which use cables like
> Analysis Plus) ?
Whose claim is this. While many things have an effect, some of those
effects are large, some are small, some probably don't exist. I
don't know anyone who claims that _everything_ makes a difference.
> Are you saying that because of the lower quality in writing audio
> data subtle differences are likely to be lost ?
Maybe, maybe not. The 16-bit resolution of the CD medium may
not be sufficient to capture the differences you are trying to
detect, particularly as you are probably not going to use ADCs
with true 16-bit performance. Or the problems in people's CD
players, which never perform as perfect 16-bit devices, may
obscure real but very small differences. As was explained to you
by another poster, before you send out your discs you need to
experiment with differences that are known to be audible and see
if they survive the coding and playback. If they don't, then your
proposed test is meaningless.
> What about greater differences like CD vs MD or different
> capacitances of the phono stage ? What about the Sony walkman vs
> Wadia scenario ?
The same points apply. You are first obliged to verify your test
procedure before you can go ahead with the test.
> This could indeed be a valid technical reason that speaks
> against such a test. However if someone manages to score
> 100%, would this prove something ?
Yes, it would show that the difference was sufficiently large
that it survived the possible problems with your experimental
procedure. (That is, by 100%, you mean scoring a high enough
number of hits to be statistically significant, like 15 out of
15. 3 out of 3 does not prove anything.)
Conversely, someone scoring 0 on your test does not prove that
the difference was inaudible. It might be audible, it might not be.
All a null result from your test would prove would be that under
the specific conditions of your test, no difference could be heard.
This is an important point but one that gets lost on many of those
who organize blind tests to "prove" that something is inaudible.
John Atkinson
Editor, Stereophile
> As was explained to you by another poster, before you send out your > discs you need to experiment with differences that are known to be > audible and see if they survive the coding and playback. If they > don't, then your proposed test is meaningless.
These differences known to be audible are known to be audible from sighted listening tests, I suppose. I really don't think that we can use results from sighted tests as reference.Furtermore I would consider as really bad luck that only those bits and bytes which express a difference between, e.g., cables are lost in the recording process.
Apart from that, if such differences are not audible on the CD this could mean both that the recording process does not get them onto the CD and that they are indeed not audible.
The Tascam uses 24bit converters so I suppose, layman that I am, that this issue is not a problem.
> These differences known to be audible are known to be audible from
> sighted listening tests, I suppose. I really don't think that we can
> use results from sighted tests as reference.
No. Before you even start your test, you have to verifyb that its
methodology is sensitive enough. I would do a dry run using something
like a 0.5dB amplitude difference, which prior work confirms is
audible. If your test methodology returns a null result with that
difference, it is unlikely to be any use with more subtle
differences, such as the ones you are talking about in this thread.
Doing correctly designed blind tests that produce meaningful results
is not easy. The literature is full of poorly designed and performed
one (even by myself :-( )
John Atkinson
Editor, Stereophile
0,5 dB difference is audible, that's a figure I have seen in literature. But at which frequency, peak or valley, narrow or broad. What signals did the research use, artificial ones or music. Did they use headphones or speakers ? The listeners of such tests usually have been trained, I'm not. Where's the threshold for the "audiophilis communis" ? Where's my own threshold ? This looks like getting complicated very quickly.Let's assume that I'm able to confirm my threshold to be 1,5 dB. You then would say, yes, your test is able to reveal such "enourmous" differences, but it yet remains open whether or not it can reveal the subtle differences we are considering here. And we are stuck again.
You are saying that the differences introduced by cables, for instance, are more subtle than 0,5 dB. Some audiophiles claim to easily hear this differences, without any training. I bend my head in awe.
This post is made possible by the generous support of people like you and our sponsors: