Quality and Usability


Customer Concern Insight Problem: TMD receives ca. 5m calls per month, >1m of which are lost.
Result: frustrated customers and lost opportunities to gain customer insight! Solution: improve “customer concern completion” by

  • Gaining insight into reasons for lost calls.
  • Improving call routing/ enabling call-back.Enabling cross-checks with business.
  • “Innovation” work package contribution to DTKS consolidation of voice portals.

Key deliverables

  • Modules for rapid and semi-automatic generation of statistical language models in given domains.
  • Evaluation and implementation of transcription of unconstrained speech using statistical language models (SLMs).
  • Voice analytics modules: improved logging for speech mining and quality auditing for contact centers, feedback loop to IVR system.
  • Computation of economic effect and customer satisfaction. Guidelines for consistent DTKS “voice user interface”.


Time Frame: 02/2008 - 07/2009

T-labs Team Members: F. Metze, C.Clemens

Students: Christin Krüger, Sven Werner

Partners: T-Systems SI, UKA/ CMU, IBM, EML, Nuance.