Semi-supervised discriminative language modeling for Turkish ASR

Abstract

We present our work on semi-supervised learning of discriminative language models where the negative examples for sentences in a text corpus are generated using confusion models for Turkish at various granularities, specifically, word, subword, syllable and phone levels. We experiment with different language models and various sampling strategies to select competing hypotheses for training with a variant of the perceptron algorithm. We find that morph-based confusion models with a sample selection strategy aiming to match the error distribution of the baseline ASR system gives the best performance. We also observe that substituting half of the supervised training examples with those obtained in a semisupervised manner gives similar results.

Learning from past mistakes: improving automatic speech recognition output via noisy-clean phrase context modeling

“…our claim of rapid adaptability of the system to varying mismatched acoustic and linguistic conditions. The extreme mismatched conditions involved in our experiments supports the possibility of going one step further and training our system on artificially generated data of noisy transformations of phrases as in [35,36,38,[57][58][59]. Thus possibly eliminating the need for an ASR for training purposes.…”

“…Further, our work is different from discriminative training of acoustic [33] models and discriminative language models (DLM) [34], which are trained directly to optimize the word error rate using the reference transcripts. DLMs in particular involve optimizing, tuning, the weights of the language model with respect to the reference transcripts and are often utilized in re-ranking n-best ASR hypotheses [34][35][36][37][38]. The main distinction and advantage with our method is the NCPCM can potentially re-introduce learning from past mistakes: improving automatic speech recognition output via noisy-clean unseen or pruned-out phrases.…”

Discriminative Language Model With Part-of-speech for Mandarin Large Vocabulary Continuous Speech Recognition System

“…A comparison of various training methods for DLMs is given in [10]. Recently, semi-supervised discriminative language modeling has also been explored [11] [12] [13]. The work in [4] is the most related to our approach.…”