BAS
Bavarian Archive for Speech Signals
Age and Gender Speech Corpus - aGender

Last update 2014-03-04 - gleiche Seite in deutsch

Beschreibung

This corpus contains 65241 recordings in real home and out-door environments spoken by 945 German speakers with the aim to develop speaker age and gender classification (7 classes). Speakers called up to 7 times a speech server and delivered read and spontaneous speech (mostly read single words). The content of each recording session is:

commands
week days
times
dates
relative times ('tomorrow')
public holidays
birth dates
telephone numbers
ZIP codes
name/surname
yes/no answers

Features:

18 recordings per session
8000Hz, 16bit, PCM, little endian
total: 2.9GB, 50.3h
vocabulary size: 2515
format and distribution: SpeechDat Exchange Format
transcript: Subset of SpeechDat Conventions

This corpus has been used in the Interspeech 2010 Paralinguistics Challenge.

From the corpus documentation

Audio Examples

Recording female 33 years Telephone number
eins drei eins acht sieben
Recording male 21 years date
siebenundzwanzigster neunter 1952

Validation Report

Availability and Costs

Free available.
Age and Gender Speech Corpus - aGender
1 DVD-R (Iso 9660) + Shipping
Scientific EUR 455,00 (ELRA Members EUR 327,00)
Commercial EUR 8.255,00 (ELRA Members EUR 8.127.00)

Questions and orders direct to