Motivation
Having a sufficient amount of training data is an essential requirement for A.I. enabled speech processing. Small and medium businesses usually lack this data. Additionally, tools to utilize data augmentation are not commonly available in the German language so far. The joint project SLIK aims to provide these options and to support German-speaking small and medium businesses that want to use speech assistants but lack the required amount of training data.
Goals
SLIK will provide the tools to enable small and medium businesses to create domain-specific synthetic speech data. To achieve this, we utilize the know-how of the three project partners: Kauz GmbH provides expertise in the field of rule-based parsing and text generation. Aristech GmbH offers their speech recognition and text-to-speech software. Saarland University shares their knowledge in low-resource deep learning. The project will deliver corpora generated by data augmentation for domains relevant to small and medium businesses as well as easy-to-use tools. These will enable the creation of speech data that can be used to train speech assistants or other automatic text- and speech-processing software.
The project-duration is 24 Months (May 2022 – April 2024).
Financial support
SLIK is being supported by the German “Bundesministerium für Bildung und Forschung” (Federal Ministry of Education and Research) in the context of “KI4KMU – Erforschung, Entwicklung und Nutzung von Methoden er Künstlichen Intelligenz in KMU (KI4KMU – Research, Development and Use of A.I. Methods in Small and Medium Businesses)”.