Transducer-based language embedding for spoken language identification
This work addresses the problem of improving language identification accuracy for multilingual speech processing, though it is incremental by building on existing transducer models.
The paper tackled spoken language identification by integrating an RNN transducer into a language embedding framework to exploit both acoustic and linguistic features, resulting in relative improvements of 12% to 59% on in-domain and 16% to 24% on cross-domain datasets.
The acoustic and linguistic features are important cues for the spoken language identification (LID) task. Recent advanced LID systems mainly use acoustic features that lack the usage of explicit linguistic feature encoding. In this paper, we propose a novel transducer-based language embedding approach for LID tasks by integrating an RNN transducer model into a language embedding framework. Benefiting from the advantages of the RNN transducer's linguistic representation capability, the proposed method can exploit both phonetically-aware acoustic features and explicit linguistic features for LID tasks. Experiments were carried out on the large-scale multilingual LibriSpeech and VoxLingua107 datasets. Experimental results showed the proposed method significantly improves the performance on LID tasks with 12% to 59% and 16% to 24% relative improvement on in-domain and cross-domain datasets, respectively.