Domain language model adaptation consists in re-estimating probabilities of a baseline LM in order to better match the specifics of a given broad topic of interest. To do so, a com mon strategy is to retrieve adaptation texts from the Web based on a given domain-representative seed text. In this paper, we study how the selection of this seed text influences the adapta tion process and the performances of resulting adapted language models in automatic speech recognition. More precisely, the goal of this original study is to analyze the differences of our Web-based adaptation approach between the supervised case, in which the seed text is manually generated, and the unsuper vised case, where the seed text is given by an automatic tran script. Experiments were carried out on data sourced from a real-world use case, more specifically, videos produced for a university YouTube channel. Results show that our approach is quite robust since the unsupervised adaptation provides sim ilar performance to the supervised case in terms of the overall perplexity and word error rate.
展开▼
