Current speech recognition systems suffer from variation of voice characteristics between speakers as they are usually based on speaker independent speech models. In order to resolve this issue, adaptation methods have been developed in many state-of-the-art systems. However, information acquired over time is still lost whenever another speaker intermittentlyuses the recognition system. This work therefore develops an integratedapproach for speech and speaker recognition in order to improve theself-learning opportunities of the system. A speaker adaptation schemeis introduced. It is suited for fast short-term and detailed long-termadaptation. These adaptation profiles are then used for an efficientspeaker recognition system. The speaker identification enables thespeaker adaptation to track different speakers which results in anoptimal long-term adaptation.

GPSR Compliance The European Union's (EU) General Product Safety Regulation (GPSR) is a set of rules that requires consumer products to be safe and our obligations to ensure this. If you have any concerns about our products you can contact us on ProductSafety@springernature.com. In case Publisher is established outside the EU, the EU authorized representative is: Springer Nature Customer Service Center GmbH Europaplatz 3 69115 Heidelberg, Germany ProductSafety@springernature.com