VB calibration to improve the interface between phone recognizer and i-vector extractor
The EM training algorithm of the classical i-vector extractor is often incorrectly described as a maximum-likelihood method. The i-vector model is however intractable: the likelihood itself and the hidden-variable posteriors needed for the EM algorithm cannot be computed in closed form. We show here that the classical i-vector extractor recipe is actually a mean-field variational Bayes (VB) recipe. This theoretical VB interpretation turns out to be of further use, because it also offers an interpretation of the newer phonetic i-vector extractor recipe, thereby unifying the two flavours of extractor. More importantly, the VB interpretation is also practically useful: it suggests ways of modifying existing i-vector extractors to make them more accurate. In particular, in existing methods, the approximate VB posterior for the GMM states is fixed, while only the parameters of the generative model are adapted. Here we explore the possibility of also mildly adjusting (calibrating) those posteriors, so that they better fit the generative model.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
Study of Phonemes Confusions in Hierarchical Automatic Phoneme Recognition System
In this paper, we have analyzed the impact of confusions on the robustness of phoneme recognitions system. The confusions are detected at the pronunciation and the confusions matrices of the phoneme recognizer. The confu…
Automatic Phoneme RecognitionPhoneme RecognitionUniversal Phone Recognition with a Multilingual Allophone System
Multilingual models can improve language processing, particularly for low resource situations, by sharing parameters across languages. Multilingual acoustic models, however, generally ignore the difference between phonem…
speech-recognitionSpeech RecognitionBuilding Tailored Speech Recognizers for Japanese Speaking Assessment
This paper presents methods for building speech recognizers tailored for Japanese speaking assessment tasks. Specifically, we build a speech recognizer that outputs phonemic labels with accent markers. Although Japanese …
Acoustically Grounded Word Embeddings for Improved Acoustics-to-Word Speech Recognition
Direct acoustics-to-word (A2W) systems for end-to-end automatic speech recognition are simpler to train, and more efficient to decode with, than sub-word systems. However, A2W systems can have difficulties at training ti…
Automatic Speech RecognitionAutomatic Speech Recognition (ASR)speech-recognitionSpeech Recognition+1Fixing Errors of the Google Voice Recognizer through Phonetic Distance Metrics
Speech recognition systems for the Spanish language, such as Google's, produce errors quite frequently when used in applications of a specific domain. These errors mostly occur when recognizing words new to the recognize…
Language ModelingLanguage Modellingspeech-recognitionSpeech Recognition