Revalidation report for the RVG1 Database

Authors Florian Schiel, Katerina Louka
Affiliation   BAS Bayerisches Archiv für Sprachsignale
Institut für Phonetik
Universität München
Postal address Schellingstr. 3
D 80799 München
E-mail schiel@phonetik.uni-muenchen.de
bas@phonetik.uni-muenchen.de
Telephone +49-89-2180-2758
Fax +49-89-2800362
Corpus Version 2.0
Date 26.11.2004
Status final
Comment  
Validation Guidelines Florian Schiel: The Validation of Speech Corpora, Bastard Verlag, 2003, www.bas.uni-muenchen.de/Forschung/BITS/TP2/Cookbook

Validation results of the RVG1 Corpus:

Summary

The speech corpus of RVG1 has been validated against general principles of good practise. The validation covered completeness, formal checks and manual checks of the selected subsamples. Missing data reduces the corpus in its usability and there could occur problems in using the corpus for other applications. 

Introduction and Corpus Description

This document summarizes the results of an in-house validation of the speech corpus  RVG1 made in the year 2004 within the project 'BITS' by the Institute of Phonetics of the Ludwig-Maximilians-University Munich.  The acronym RVG1 stands for  Regional Variants of German 1.

The corpus consists of single digits, connected digits, phone numbers, phonetically balanced sentences, computer command phrases and spontaneous
speech.

I.) Validation of Documentation

The General Documentation directory contains the following documentation files for the RVG1 corpus which can be found under: doc/

README general documentation
SPRK_ATT.TXT
 speaker information
PROMPTS
 directory with prompt lists
 TRLMAN.PS
description of Verbmobil II transliteration format
VMTRLEX2D.HTML
 description of Verbmobil II transliteration format (German)
RVG1_TRL.LIST
 word list spontaneous speech
RVG1_TRL.LEX
canonic pronunciation in SAM-PA spontaneous speech
RVG1_READ.LIST
 word list read speech
RVG1_READ.LEX
canonic pronunciation in SAM-PA read speech
PARDOC
documentation of the BAS Partitur Format
SPRK_DIR.TXT
 information about the location of the data of each speaker

The main directory of each volume contains following directories:

README general documentation
A-I
 speech data of dialect regions a-i
DOC
 General Documentation directory 
SOFTW
 BAS standard software package
TRL
 Transliteration Files to spontaneous monologues
PAR Bas Partitur Files to all utterances (BPF)       (only on volumes RVG1_LQ_19 and RVG1_HQ_12)
LABELS
original validation files (only on volumes RVG1_LQ_20 and RVG1_HQ_12)

          Formats of signals and annotation files: If non standard formats are used it is common to give a full description or to convert into a standard format:

        The signal file format is NIST SPHERE.ok

        Compression: Just widely supported compressions like zip or gzip should be used.
        n. a.

        Sampling rate: 22050 Hz (001 to 036 were recorded with 11.025 Hz)ok

        Valid bits per sample: (others than 8, 16, 24, should be reported): 16 bits/samp ok

        Used bytes per sample: 2 bytes/samp ok

        Multiplexed signals: (exact de-multiplexing algorithm; tools)
        n. a.

        Database contents:

        Clearly stated purpose of the recordings:
        No information, not important for this corpus

        Speech type(s): (multi-party conversations, human-human dialogues, read sentences, connected and/or isolated digits, isolated words etc.) ok

        Instruction to speakers in full copy: no information not ok

        Linguistic contents of prompted speech:

        Specifications of the individual text items: ok

        Specification for the prompt sheet design or specification of the design of the speech prompts:  not given, not ok

        Example prompt sheet or example sound file from the speech prompting: not given, not ok

        Linguistic contents of non-prompted speech:

        Multi-party:(number of speakers, topics discussed, type of setting - formal/informal) n. a.

        Human-human dialogues: (type of dialogues, e.g. problem solving, information seeking, chat etc., relation between speakers, topic(s) discussed, type of         setting, scenarios) n. a.

        Human-machine dialogues: (domain(s), topic(s), dialogues strategy followed by the machine, e.g. system driven, mixed initiative, type of system, e.g.             test, operational service, Wizard-of-Oz) n.a.

Speaker information:

Speaker recruitment strategies: No information, not importatnt for this corpus

Number of speakers: 501 ( 282 male speakers and 219 female) ok

Coding: PCM linear ok

           Distribution of speakers over sex, age, dialect regions: given in  the file "sprk_att.txt" and in the file "dialects.txt" ok
           Description/definition of dialect regions:  given in the README file

          Status of documentation: acceptable

II.) Automatic validation

The following list contains all validation steps with the methodology and results.

Completeness of signal files:

           - 011
           - 146
           - 162
           - 283
           - 316
 

Completeness of meta data files: ok

Completeness of annotation files: 

          - 145
          - 146
          - 316
          - 145: No spontaneaus speech par file
          - 146: No spontaneaus speech par file
          - 316: Some phone number with area code par files are missing, no spontaneaus speech par file
          - 330: Exists only the spontaneaus speech par file
          - 331: Exists only the spontaneaus speech par file
          - 331: Exists only the spontaneaus speech par file
          - 332: Exists only the spontaneaus speech par file
          - 352: Phone number with area code files are missing (2)
          - There are no validations files for the speakers 330-331-332
          - 352: 2 phone number with area code  validation files are missing
          - 146: No spontaneaus speech validation file
          - 316: 2 phone number with area code validation files are missing, no spontaneaus speech validation file

Correctness of file names: ok.

Empty files: none

Status of signal, annotation and meta data files: ok

Cross checks of meta information: ok

Cross checks of summary listings: ok

Annotation and lexicon contents: Following words are differently annoted in the par files as in the lexicon files:

III.) Manual Validation

5% of the data and annotations files was checked in comparison. 2,65% of the data contained errors. The most errors were found in
annotation of the spontaneaus speech files.

IV.) Other Relevant Observations

none

V.) Comments for Improvement

The revalidation was able to repair some data and to add some more informations about the speakers' data . The results of the manual validation
couldn't be repaired.

VI.) Result

The corpus is ok. No documentations files are missing and  the corpus is well documented.