OLLO Banner

Oldenburg Logatome Corpus

The Oldenburg Logatome Corpus (OLLO) is a speech database that was designed for a comparison of recognition performance of automatic speech recognizers and human listeners. It has also been useful for other purposes, e.g., the development of a speaker discrimination task, or modelling phoneme recognition.

OLLO contains simple non-sense combinations of consonants (C) and vowels (V), which are referred to as logatomes. 150 different CVCs and VCVs were spoken by 40 German and 10 French speakers. Each logatome was recorded in several different speaking styles (loud and soft speaking effort, fast and slow speaking rate, or with rising pitch), which enables an analysis of these speech-intrinsic variabilities. Additionally, it may be used to analyze the influence of dialect and accent, as speakers from different dialect regions as well as French speakers have been recorded.

OLLO has been described in detail in Meyer et al., J. Acoust. Soc. Am.128 (5), 3126-3141. A comparison of HSR and ASR data is presented in Meyer et al., J. Acoust. Soc. Am 129, 388-403. You can visit this homepage to obtain a free copy of these articles. More articles that use the database are listed on the reference page.


man-machine comparison

Examples for the logatome 'ollo', spoken normally, followed by variabilities 'fast', 'slow', 'loud', 'soft', and 'question'

[download wave file]

Examples for the same logatome spoken by male speakers without dialect, followed from dialect speakers from East Frisia, Bavaria, East Phalia, and France

[Download example #1]

[Download example #2]






Last update: 21-Apr-2009 5:07 PM