HMM-based Non-native Accent Assessment using Posterior Features

Automatic non-native accent assessment has many potential benefits in language learning and speech technologies. The three fundamental challenges in automatic accent assessment are to characterize, model and assess individual variation in speech of the non-native speaker. In our recent work, accentedness score was automatically obtained by comparing two phone probability sequences obtained through instances of non-native and native speech. In this paper, we build on the previous work and obtain the native latent symbol probability sequence through the word hypothesis modeled as a hidden Markov model (HMM). The approach overcomes the necessity for a native human reference speech of the same sentence. Using the HMMs trained on an auxiliary native speech corpus, the proposed approach achieves a correlation of 0.68 with the human accent ratings on the ISLE corpus. This is further interesting considering that the approach does not use any non-native data and human accent ratings at any stage of the system development.

Related material