Using of heterogeneous corpora for training of an ASR system
This work addresses the problem of improving speech recognition for low-resource languages like Pashto, where data heterogeneity complicates training, but it appears incremental as it focuses on adapting existing techniques rather than introducing a new paradigm.
The paper tackled the challenge of training an ASR system for Pashto, a low-resource language, using heterogeneous corpora from multiple sources, where preliminary experiments showed that simply merging the data provided little to no benefit, and it presented various techniques to effectively utilize all the data.
The paper summarizes the development of the LVCSR system built as a part of the Pashto speech-translation system at the SCALE (Summer Camp for Applied Language Exploration) 2015 workshop on "Speech-to-text-translation for low-resource languages". The Pashto language was chosen as a good "proxy" low-resource language, exhibiting multiple phenomena which make the speech-recognition and and speech-to-text-translation systems development hard. Even when the amount of data is seemingly sufficient, given the fact that the data originates from multiple sources, the preliminary experiments reveal that there is little to no benefit in merging (concatenating) the corpora and more elaborate ways of making use of all of the data must be worked out. This paper concentrates only on the LVCSR part and presents a range of different techniques that were found to be useful in order to benefit from multiple different corpora