Adapt SHAWI data management workflows to include audio files

[simar0at]

We have wav files and we have timestamps in TEI XML.

We need a way to cut the wav files and also probably to export them as mp4.

One way to do this is to transform the TEI files to audacity 2 project files which also happen to be XML files.

[simar0at]

We can generate the labels in text format:

0.000000	0.439679	H
0.346413	0.912666	IG
0.925990	2.394558	IG
2.394558	5.009671	IG
2.711361	2.851247	H
5.006803	5.558277	ID
5.558277	8.576871	IG
8.576871	11.116553	IG
11.116553	23.942498	H

We can use this with audacity 2 or 3. Audacity 3 moved away from any (visible) XML.

[dasch124]

IMO we should go for the CSV/TSV format which seems less effort to create.

[dasch124]

General Workflow:

1. For each TEI transcription document, we generate "region labels" (i.e. named time spans in Audacity) in the format mentioned by Omar in the comment above and add them to the SHAWI data repository.
2. Team members at university with access to the original audio files open the files in Audacity, import the region label list and export the audio snippets both as uncompressed WAV (for archiving) and MP3 (for publishing in application)
3. They upload the resulting audio files to https://oeawcloud.oeaw.ac.at/index.php/apps/files/?dir=/R_Shawi_19367&fileid=36169042
4. We add references to both versions of the audio files to the TEI documents.

TEI > Auacity labels conversion

The TSV format is described here: https://manual.audacityteam.org/man/importing_and_exporting_labels.html

This should be generated by taking all <u> elements in the transcription documents, and re-calculating the absolute timestamps from the @interval attribute on the <when> elements inside of the <timeline>:

  <timeline unit="ms">
         <when xml:id="T0"/>
         …
         <when interval="197124" since="#T0" xml:id="T19"/>
         <when interval="197256" since="#T0" xml:id="T20"/>
          …
      </timeline>
     …
    <annotationBlock>
               <u xml:lang="ar-acm-x-shawi-vicav" xml:id="URFA-034_a20" who="#default" end="#T20" start="#T0">
                  … 
              </u>
    <annotationBlock>    

Instead of having the speaker name as the label, we should use the utterance's xml:id, so the exported audio snippet can be named after the utterance id.

[dasch124]

for some reason, the xml:id is missing from the @url attribute on , e.g. https://github.com/acdh-oeaw/shawi-data/blob/main/010_manannot/Urfa-097_Three_Daughters-Harran-2010.xml#L210

[rausch-supola]

As I said I only inserted the two lines with the media tag. I guess the linking to some data is missing

[rausch-supola]

this issue can be closed I guess @dasch124 ?