BAS Partitur Format files (BPF) TNNN_SD.par

The BAS Partitur Format is a simple but effective way to represent 
symbolic (discrete) labels (categories together with their time
information) aligned to a physical signal. Developed for the 
multi-tier annotation of speech signals it was also used in the 
SmartKom project to summarize and align all the annotation layers
(tiers) associated with different (time synchronized) signals.

The main (and up-to-date) documentation can be found in 
www.bas.uni-muenchen.de/Bas/BasFormatseng.html
(a copy of this page at the time of distribution can be found 
in 'doc/pardoc/index.html').

This README.PAR does only contain some general remarks about the 
usage of BPF within SmartKom. For the details of the individual 
tiers please refer to the main documentation. For a more general 
description of the concept please refer to 
'doc/papers/Schiel-02-LREC-WS.ps'.


BPF is a format that uses two ways to align different information 
tiers in a recording: the physical time scale and a discrete numbering
of events that may be associated to most of the tiers: the word order
(numbered beginning with 0).
For the SmartKom corpus this has the following consequences:

- To each recording there exist two BPF files, e.g.
  w157_mn_ADA.par contains the BPF of the speaker 'ADA' and 
  w157_mn_SMA.par contains the BPF of the SmartKom system 
  as they communicate in the recording session 'w157_mn'
  (the SmartKom system has always the ID 'SMA')
- Each BPF file contains all events and annotations for the whole 
  session; time is measured from the beginning of the time synchronized
  signals files. If not given otherwise the unit of time is 1/16000 sec.
- the following tiers are given in a SK BPF file: 
  TRS : basic transliteration broken into word units
  TRN : beginning and duration of turns numbered as in the transliteration
        format
  ORT : orthography as used in the pronunciation dictionary
  KAN : canonic pronunciation as given in the pronunciation dictionary
  SUP : crosstalk between speaker and system
  NOI : noise markers associated to words or the gaps between words
  GES : 2-D gesture labeling with reference to the time scale
  USH : holistic user state labeling and segmentation with reference 
        to the time scale
  USM : user state labeling and segmentation from facial expression 
        with reference to the time scale
  USP : meta-linguistic labeling and segmentation from voice input
        with reference to words and to the time scale as input to 
	automatic user state detection from voice input
  MAU : automatic phonetic/phonemic segmentation and labelling
        by MAUS 
  	
Since the BPF files integrate all other annotations into a single
format and since these annotations will be error pruned and/or 
updated from time to time, it is clear that the update rate of the 
BPFs are quite high (every time a annotation is getting updated).
Therefore, we strongly recommend that you download the newest 
version of the BPF files before any critical work with the SK data.

Updated versions of the BPF files of SmartKom may be downloaded from
ftp://ftp.bas.uni-muenchen.de/pub/BAS/SK