audio technical metadata

[jody]

Hi --

    I'm analyzing the AES57-2011 (AES standard for audio metadata - Audio object structures for preservation and restoration -- schema here) to map our technical metadata, and I find I have questions that the standard documentation does not answer. I'm wondering if anyone here is deeply familiar with audio technical metadata.

 

   My questions for today (yes, I have others!) are about start time and duration. In the above schema, the "timeline" element is used to indicate the duration and the timestamp of the start of audio for the enclosing face element (for example, one side of a phonograph record).  The timeline element contains two elements, "startTime" and "duration", each of type LONG, and there is NO information as to what to use as a type of measurement.  This is where my confusion comes in.  

It seems that much of this standard was build on the work described in "Sound Directions:  Best Practices for Audio Preservation" (by Mike Casey, Indiana U and Bruce Gordon, Harvard) -- in that document, start times are always of the form h:mm:ss, and duration is of the number of frames.  

  So:  how exactly should one express a time stamp as type LONG?    Or should I be identifying how many frames occur (0?) before the audio begins?

  And secondly, where does one obtain these counts of frames, particularly for regions within the face?  (There must be some tools for this;  or can one calculate the number of frames per second?  If so, where would I find that formula?)

  Third:  in the FITS (File Information Tool Set, Harvard) output for a typical WAV file, the "OIS Audio Information" tool (maintained by Harvard, but about which I can find no documentation whatsoever) provides a numSamples value (which is a LONG) and also a time value (which is also a LONG), and these two differ.   I would guess that the numSamples *is* the number of frames in the WAV file, but I don't know how to verify that.  I have no idea what measurement is used for the time value.  Here's an example, and I will include two duration measures by other tools for the same file, as I can find no relationship between the duration and these other numbers:

      <numSamples toolname="OIS Audio Information" toolversion="0.1" status="SINGLE_RESULT">55881728</numSamples>

      <sampleRate toolname="OIS Audio Information" toolversion="0.1">44100</sampleRate>

      <time toolname="OIS Audio Information" toolversion="0.1" status="SINGLE_RESULT">1919849400</time>

      <duration toolname="Exiftool" toolversion="9.13" status="CONFLICT">0:21:07</duration>

      <duration toolname="NLNZ Metadata Extractor" toolversion="3.4GA" status="CONFLICT">0:21:7:166</duration>

I'm wondering if I should use "numSamples" or "time" as duration, and simply use "0" for timestamp -- and then forget about region information.

Can anyone here help clear up my confusion and point me to useful documentation?  I would very much like to capture and store useful technical metadata for our audio, and it would be nice if I could use a single good standard for it.

Thank you!

Jody DeRidder

Head, Digital Services

University of Alabama Libraries

[jody]

Update:  I have partially answered my own questions, for those of you who are interested.

numSamples /  sampleRate = seconds.

So I can multiply the number of seconds times the sample rate to get a number (of type LONG) for the startTime and for duration, for the face, as well as for each region within the face.   I still have no clue as to what the "time" field is that is measured by "OIS Audio Information" in the FITS toolkit, but hey, I now know enough to make some progress in filling in the fields of the AES standard.

Anyone interested in this topic and other aspects of it -- please email me, and we can continue this discussion offlist.

Thanks!

--jody

jlder...@ua.edu

[David Ackerman]

Hi Jody,

Here is a timeline entry for a real file from a tool I have been working on that implements AES57. 

<timeRange>

                <startTime editRate="44100" factorNumerator="1" factorDenominator="1">0</startTime>

                <duration editRate="44100" factorNumerator="1" factorDenominator="1">1031935</duration>

</timeRange>

The startTime and the duration elements are borrowed from AES60/EBUCore's editUnitNumberType. Their documentation for that reads:

A number of edit units. An editUnit is the inverse of the edit rate, or corrected edit rate as the result of editUnit=1/(editrate*(factorNumerator/factorDenominator)) 

So, in this case a single edit unit is 1/(44100*(1/1) = 0.00002267573696

the start time is 0 and the duration is 1031935 *  0.00002267573696 = 23.3998866198176 seconds

The factorNumerator and factorDenominator are useful for expressing values with a pulldown or pull up applied. For example, NTSC video runs at 1000/1001 to account for the 1% slowdown applied. 

e.g. <startTime editRate="30" factorNumerator="1000" factorDenominator="1001">0</startTime>

The start address install zero but the time units are now NTSC frames. 1/(30*(1000/1001)) = 29.97002997003 FPS

So it is really a flexible system for describing time in whatever units make sense for your application. In the first case it is samples and in the second NTSC video frames.

-David

[Jody L. DeRidder]

David, thank you so much!!

 

---
Jody L. DeRidder
Head, Digital Services
University of Alabama Libraries
Tuscaloosa, AL 35487
Phone: 205.348.0511


"Hope lies in dreams, in imagination, and in the courage of those who dare to make dreams into reality." --Jonas Salk

--
You received this message because you are subscribed to the Google Groups "Digital Curation" group.
To unsubscribe from this group and stop receiving emails from it, send an email to digital-curati...@googlegroups.com.
To post to this group, send email to digital-...@googlegroups.com.
Visit this group at http://groups.google.com/group/digital-curation.
For more options, visit https://groups.google.com/d/optout.