[MEI-L] Time representation in MEI and EDTF

[Byrd, Donald A.]

On Fri, 05 Apr 2013 08:41:22 +0200, Benjamin Wolff Bohl 
<bohl at edirom.de> wrote:
> Hi Don,
> took me a while to look what this interesting email subject was all
> about and it is really interesting! The notions of implementing
> certainty in a timeformat that EDTF puts forward is certainly very
> promising for dating issues, although this might be conflicting with
> the possibilities metadata formats already give. An implementation of
> a fixed EDTF standard in a format already providing possibilities
> like certainty/uncertainty, not-before, not-after, etc. will result
> in even more ways of encoding the same thing, as metadata formats
> certainly won't easily drop such features.
> If it comes to music the decimal digits of seconds might not be
> interesting for source dating but for metadata of audio files, CDs,
> LPs or any other sound carrier. Also in things like performance
> analysis, or semiautomatic audio data analysis or some MIR issues,
> precise timecodes are essential.
> From my audio engineering background I'd even say that you'll need at
> least 3 digits decimal places. In binaural tests 10ms delay between
> the same signal on both ears have been proved to be audible by almost
> everybody. Trained audio engineers / musicians (the best of wich were
> percussionists) can hear differences smaller than 5ms, even a 1ms
> delay can be heard.

Hi, Benjamin. Thanks for the reply; _I_ thought this was an interesting 
question, and one really deserving discussion! What you say about 
audible time differences agrees closely with what I've read and my own 
(limited) experience. Though I'm not convinced yet anyone can hear a 1 
ms delay -- I would have said 2 ms or so -- but you probably know more 
than I do.


> Now for example describing digitized audio material, I don't see why
> one should not be able to describe sample positions, e.g.
> hh:mm:ss:sample.
> With audio samplerates of standardized sound carriers ranging as high
> as 192kHz this would mean a sample is something like 0.0052
> milliseconds!
> Nevertheless there are phenomena on intersample level which lead DAW
> (Digital Audio Workstation) plugin developers to start oversampling
> in their software.
> The more technical we get on this the smaller will the time units get ;-)

:-)  Exactly. Oversampling involves frequencies in the megahertz, so 
intervals of less than a microsecond. But I think the main issues for 
EDTF are:

(1) What range of applications it's intended to support. The official 
statement is the effort is "to develop a reasonably comprehensive 
date/time definition for the bibliographic community, as well as other 
interested communities, and submitting it for standardization or some 
other mode of formalization, for example a W3C note or an amendment to 
ISO 8601." Well, what other communities are interested? I suspect the 
academic music-encoding community is, and you and I (at least) agree 
that resolution of one second isn't nearly good enough!

(2) Compatibility. In particular, is EDTF supposed to have the 
potential to replace ISO 8601? To a great extent, this is a tradeoff 
with difficulty of implementation. The list moderator's position seems 
to be "not if it makes implementing EDTF harder", but I think that's a 
mistake. Especially since they've already divided features into several 
levels, i.e., Level 0 is a subset of ISO 8601; Level 1 adds many 
features, and Level 2 adds more. So people that want only what ISO 8601 
can deliver need only support EDTF Level 0.

I think the major issues for MEI are similar. We certainly want to 
describe audio timing as accurately as anyone can hear, and we may want 
to describe digitized audio at the sample level -- but not considering 
oversampling, which is really just a convenience for the D/A converter. 
And we want to maintain compatibility with existing time 
representations _if_ it's not too hard.

There's a lot more to say about all of this, but I'd better stop with 
the above twenty-two cents :-) .

--DAB


> To cite someone fequently posting here:
> "just my two cents"
>
> cheers,
> Benjamin
>
> Am 26.02.2013 04:22, schrieb Byrd, Donald A.:
>> All--
>>
>> I've been thinking a lot recently about representing time in
>> extremely general ways, and I've gotten involved in discussion of
>> EDTF, an "Extended Data Time Format" under development by the
>> bibliographic community (http://www.loc.gov/standards/datetime/). In
>> many ways, EDTF is "extended" compared to, say, ISO 8601, but the
>> current draft is actually more limited in at least one important
>> way: its maximum precision is a second, while both 8601 and XSD
>> allow seconds with several -- quite possibly an unlimited number of
>> -- decimal places.
>>
>> I'm telling you this because I'd like to see a musicologist or two
>> involved in the EDTF discussion. Furthermore, I'd like to see
>> support for seconds with multiple decimal places added to EDTF, and
>> the EDTF coordinator at the Library of Congress seems willing to add
>> it _if_ someone gives him a realistic use case. I'm not sure I have
>> a realistic bibliographic use case, but I'll bet some of you do!
>>
>> Also, a side question. The MEI docs keep referring to "standard ISO
>> form" for dates and times. What is "standard ISO form"? I can't find
>> an explanation. Does that mean ISO 8601?
>>
>> --Don
>>
>>
>>
>> _______________________________________________
>> mei-l mailing list
>> mei-l at lists.uni-paderborn.de
>> https://lists.uni-paderborn.de/mailman/listinfo/mei-l
>
>
> _______________________________________________
> mei-l mailing list
> mei-l at lists.uni-paderborn.de
> https://lists.uni-paderborn.de/mailman/listinfo/mei-l
>


--
Donald Byrd
Woodrow Wilson Indiana Teaching Fellow
Adjunct Associate Professor of Informatics
Visiting Scientist, Research Technologies
Indiana University Bloomington