A Broadcast News Corpus for Evaluation and Tuning of German LVCSR Systems
This work provides a domain-specific evaluation framework for German LVCSR systems, focusing on broadcast news transcription, and is incremental as it applies existing methods to new data.
The authors tackled the problem of evaluating and tuning German large-vocabulary continuous speech recognition (LVCSR) systems by creating a manually segmented and annotated corpus of over 160 hours of German broadcast news. They demonstrated that real-time decoding on this test set is feasible on a desktop PC, achieving a word error rate of 9.2%.
Transcription of broadcast news is an interesting and challenging application for large-vocabulary continuous speech recognition (LVCSR). We present in detail the structure of a manually segmented and annotated corpus including over 160 hours of German broadcast news, and propose it as an evaluation framework of LVCSR systems. We show our own experimental results on the corpus, achieved with a state-of-the-art LVCSR decoder, measuring the effect of different feature sets and decoding parameters, and thereby demonstrate that real-time decoding of our test set is feasible on a desktop PC at 9.2% word error rate.