SUB-COMMITTEE ON SUB-SAHARAN AFRICA
ISCA International Affairs Committee
Report 2008
Status
This is the fourth report of the Sub-committee, prepared by the Co-coordinator. Members of the sub-committee are:
§ Prof Justus C Roux () , Director: Stellenbosch University Centre for Language and Speech Technology, South Africa. (Coordinator)
§ Dr. Tunde Adegbola (, , Associate lecturer, University of Ibadan, Ibadan (Nigeria) (also Executive Director of the African Languages Technology Initiative)
§ Dr. Nimaan Abdillahi (), Director, Djibouti Institute of Science and Information Technologies, Djibouti
Activities
It is not possible to report on each and every project that is taking place in the region, however, some of the main known institutions within this area (South, West and East Africa) are listed below with some information on active projects. (Detail to be obtained from websites listed.)
West Africa
Nigeria: African Language Technology Initiative (ALT-i), Nigeria, (http://www.alt-i.org)
Corpus Development and Language Modelling Yoruba
Yoruba Keyboard and Word processor development
Speech recognition and TTS for Yoruba
Igbo-English Machine Translation
In addressing other Nigerian languages, an active working relationship exist with the National Institute for Nigerian Languages (NINLAN), in Aba, Nigeria
East Africa
Djibouti: Djibouti Centre of Speech Research, Djibouti
ASR on AFAR language (the second language of Djibouti)
Automatic translation between Somali-French-Afar
Kenya: Teknobyte Speech Technologies, (www.teknobyte.co.ke)
Speech synthesis in Kiswahili
Lead consultant in the development of NAFIS, the first ever automated IVR for use by rural farmers in Kenya. NAFIS uses Teknobyte's Kiswahili and Kenyan English voices, and has been developed in collaboration with Speechnet Ltd for the Ministry of Agriculture and Ministry of Livestock development (cf www.nafis.go.ke )
South Africa
The following Centres / Departments are actively involved in speech processing research and development (detail on websites):
(i) Meraka Institute, Pretoria, South Africa (http://www.meraka.org.za)
The following projects may be listed:
Lwazi:
Development of tools and technologies that support the provision of multilingual telephony-based information services in all South Africa's official languages.
§ the creation of an Open Content linguistic resource repository containing annotated speech data for ASR and TTS purposes in 11 languages
§ the development of ASR & TTS systems in 11 languages
§ all data and tools are distributed free of charge under an Open Source license.
OpenPhone
Development, piloting and evaluation of a telephony-based information service capable of assisting home caregivers in caring for HIV-positive children in Gaborone, Botswana.
§ an ASR and TTS system in the dialect of Setswana spoken in the Gaborone region.
PAST (Phonetics for Advanced Speech Technology)
A collaborative project – Universities of Witwatersrand and North-West - focus on important linguistic questions for the development of advanced speech technology in South Africa. Include questions related to phonetic and phonological systems, the prevalence and influence of dialects, and the modelling of tone.
(ii) Stellenbosch University Centre for Language and Speech Technology (SU-CLaST), Stellenbosch, South Africa (http://www.sun.ac.za/su_clast )
§ Speech processing
– Speech-to-Speech Translation
– High quality speech synthesis in African tone languages (with ATR, Japan)
– Large vocabulary speech recognition for SA English
– Multilingual acoustic modelling.
– Multilingual and multi-accent speech recognition in SA languages
§ Mobile learning systems
– Implementing multilingual TTS based learning systems on mobile platforms in collaboration with University of Copenhagen
(iii) DSP Laboratory, Stellenbosch University, Stellenbosch, South Africa (http://www.ee.sun.ac.za/research/signal_processing.php)
§ Speech processing
Interactive Naval Reference dictionary
Afrilator: An English-Afrikaans speech translator
English-Xhosa Speech translator
§ Speaker identification
(iv) University of Limpopo Telkom Centre of Excellence for Speech Technology, South Africa (http://www.unorth.ac.za/facultySchools/sch-comp/disciplines/comp_sci/coe/index.html)
§ Development of speech synthesizers for African languages: N Sotho (Sepedi), Tswana, Venda and Tsonga;
§ Tools for automatic recognition of continuous speech for N Sotho (Sepedi), Tswana, Venda and Tsonga;
(v) University of Cape Town Speech Technology and Research Group, Cape Town, South Africa (http://www.star.za.net)
§ Focus on speaker recognition.
Conferences / workshops
ODYSSEY 2008 (http://www.speakerodyssey.com).
An ISCA supported Odyssey Workshop on Speaker and Language Recognition took take place in Stellenbosch, South Africa from 21 to 25 January 2008. This workshop was well attended and very successful. (Contact: Niko Brummer; )
This follows the successful presentation of another ISCA workshop Multiling 2006 that was also presented at Stellenbosch and mentioned in the previous Sub-committee Report.
Conference on Human Language Technologies in Africa 2008
This conference was sponsored by the Alexander von Humboldt Stiftung (Germany) and took place in Rabat (Morocco) from 3-5 June 2008. It brought together researchers from various areas within the whole of Africa (also with European participation). See Appendix A.
Participation in the 6th ISCA Speech Synthesis Workshop, Bonn (22-24 August 2007)
Presentation on TTS in African languages: cf. Roux, JC & AS Visagie: Data-driven approach to rapid prototyping Xhosa speech synthesis.
Annual Conference of the Pattern Recognition Association of South Africa (PRASA)
The annual PRASA conferences make provision for topics on speech processing (see http://www.prasa.org)
(Note: it was not possible to determine participation of other colleagues in national and international conferences. Details should be available on respective websites.)
ISCA Distinguished Lecturer
Prof Marc Swerts visited South Africa for ten days in July 2008. He gave a talk at Meraka-Institute entitled “Facial cues to information status”, and presented a keynote address at the Interim Conference of the African Languages Association for Sothern Africa (ALASA) in Swaziland. He also made a presentation at the University of the Witwatersrand in Johannesburg.
Justus Roux Stellenbosch
Co-ordinator: Sub-committee of Sub-saharan Africa 4 August 2007
/isca/sub_com_rep4
APPENDIX A
REPORT of the Alexander von Humboldt Kolleg “Human Language Technologies in Africa: Status and Prospects”
June 3-5, 2008, MOROCCO
Prof. Abdelhadi Soudi and Prof. Justus Roux)
The Ecole Nationale de l’Industrie Minérale in Rabat (Morocco) hosted the Alexander von Humboldt Kolleg on “Human Language Technologies in Africa: Status and Prospects”, 3-5 June 2008.
The Kolleg aimed at developing stronger research ties between Humboldtians around the world and particularly within Africa, with some key actors in Germany and junior scientists working on Human Language Technologies (HLT).
The Kolleg was attended by 27 persons from 10 countries (10 Humboldtians, 2 German senior scientists, 8 junior scientists and 7 other participants). 24 papers were presented (cf. attached programme).
The keynote address by Prof Justus Roux of Stellenbosch (South Africa) aimed at setting the scene and providing background to the topic of the conference. The title of the paper was: “Human Language Technologies in Africa: Needs and Challenges”.
Following this presentation, the status and prospects for HLT was discussed through a number of presentations during which the following key questions were addressed:
· Issues in the design, the acquisition, creation, management, access, distribution, use of Language Resources (LRs);
· Exploitation of LRs in different types of applications (machine translation, Information Retrieval, Question-Answering etc.;
· Ways of safeguarding linguistic and cultural diversity in the information society of tomorrow by strengthening the position of the languages used in Africa in linguistics and language technology;
· Human Language Technologies as a means to assist industry and government to make services and documents available in the languages (spoken and written) in Africa;
· Human Language Technologies’ role in rectifying the historical discrimination against specific languages;
· Sharing of methods and data;
During a final round table discussion, the AvH Kolleg took the following decisions for further action and sustainability of the collaboration:
(i) To approach a reputable publishing house (Springer Verlag) to publish extended presentations of these presentations to make the views expressed to a wider audience
(ii) To establish an “International Group for Human Language Technologies in Africa” as a first step towards an international network to specify and support the development of high priority Language Resources for the languages in Africa in a systematic, standards-driven, collaborative learning context. The lead of the group will be initially taken by Prof. Justus Roux (university of Stellenbosch, South Africa), Prof. Abdelhadi (Ecole Nationale de l’Industrie Minérale, Morocco), Prof. Firmin Ahoua (University of Cocody, Ivory Coast), Prof. Dafydd Gibbon (Universität Bielefeld, Germany), Prof. Günter Neumann (DFKI, Germany). This Group will advance its activities in two phases:
Phase 1
Identification of the state of the art of Language Resources in Africa, assessing priority requirements through consultations with language industry and communication players
Phase 2:
Promote active collaboration in the field in Africa, inter alia, through joint research, LR and application development and training activities. It will also be the task of the management of this group to leverage international funding for these activities.
The organisers wish to thank the AvH Stiftung for the generous support for this event that has opened new horizons for HLT development and deployment in Africa. This will contribute to keep Africa abreast with developments of the Information Society, and assist in bridging the ever growing digital divide.
6