A Comparison of Hybrid and End-to-End Models for Syllable Recognition
This work addresses speech recognition for syllabic or agglutinative languages, offering insights for tasks like keyword search and medical speech processing, but it is incremental as it compares existing methods.
The paper compared hybrid and end-to-end models for German syllable recognition, finding that a hybrid system with a 4-gram language model achieved 10.0% word error rate, significantly outperforming the end-to-end model's 27.53%.
This paper presents a comparison of a traditional hybrid speech recognition system (kaldi using WFST and TDNN with lattice-free MMI) and a lexicon-free end-to-end (TensorFlow implementation of multi-layer LSTM with CTC training) models for German syllable recognition on the Verbmobil corpus. The results show that explicitly modeling prior knowledge is still valuable in building recognition systems. With a strong language model (LM) based on syllables, the structured approach significantly outperforms the end-to-end model. The best word error rate (WER) regarding syllables was achieved using kaldi with a 4-gram LM, modeling all syllables observed in the training set. It achieved 10.0% WER w.r.t. the syllables, compared to the end-to-end approach where the best WER was 27.53%. The work presented here has implications for building future recognition systems that operate independent of a large vocabulary, as typically used in a tasks such as recognition of syllabic or agglutinative languages, out-of-vocabulary techniques, keyword search indexing and medical speech processing.