In this work we aim to discover high quality speech features and linguistic units directly from unlabeled speech data in a zero resource scenario. The results are evaluated using the metrics and corpora proposed in the Zero Resource Speech Challenge organized at Interspeech 2015. A Multi-layered Acoustic Tokenizer (MAT) was proposed for automatic discovery of multiple sets of acoustic tokens from the given corpus. Each acoustic token set is specified by a set of hyperparameters that describe the model configuration. These sets of acoustic tokens carry different characteristics fof the given corpus and the language behind, thus can be mutually reinforced. The multiple sets of token labels are then used as the targets of a Multi-target Deep Neural Network (MDNN) trained on low-level acoustic features. Bottleneck features extracted from the MDNN are then used as the feedback input to the MAT and the MDNN itself in the next iteration. We call this iterative deep learning framework the Multi-layered Acoustic Tokenizing Deep Neural Network (MAT-DNN), which generates both high quality speech features for the Track 1 of the Challenge and acoustic tokens for the Track 2 of the Challenge. In addition, we performed extra experiments on the same corpora on the application of query-by-example spoken term detection. The experimental results showed the iterative deep learning framework of MAT-DNN improved the detection performance due to better underlying speech features and acoustic tokens.

Silicon nitride (SiN(x)) and parylene thin films were deposited onto flexible polyimide (PI) substrates using plasma-enhanced chemical vapor deposition and a parylene reactor for transparent barrier applications. The PI substrates from the Industry Technology Research Institute with high optical transmittance and high glass transition temperature were used. A relatively high growth temperature of 200 degrees C was chosen to deposit the SiN(x) films. To characterize the SiN(x) films deposited under different growth temperatures, a wet-etching process was performed to visualize the defect distribution in the barrier films. After 120 min of etching, the etching area ratio decreased from 44.9 to 6.7%, while the average defect spacing increased from 125 to 450 mu m with increasing growth temperature. Under room temperature and relative humidity of 50%, four SiN(x)/parylene stacks with the SiN(x) films deposited at 80 and 200 degrees C were demonstrated to decrease the water vapor transmission rate to 7.9x10(-4) and 7.41x10(-6) g/m(2)/day, respectively. As a result, ultralow permeation can be achieved with less repeating barrier stacks by using high temperature deposited SiN(x) films in the barrier structures

Abstract-In this paper we aim to automatically discover high quality frame-level speech features and acoustic tokens directly from unlabeled speech data. A Multi-granular Acoustic Tokenizer (MAT) was proposed for automatic discovery of multiple sets of acoustic tokens from the given corpus. Each acoustic token set is specified by a set of hyperparameters describing the model configuration. These different sets of acoustic tokens carry different characteristics for the given corpus and the language behind, thus can be mutually reinforced. The multiple sets of token labels are then used as the targets of a Multi-target Deep Neural Network (MDNN) trained on frame-level acoustic features. Bottleneck features extracted from the MDNN are then used as the feedback input to the MAT and the MDNN itself in the next iteration. The multi-granular acoustic token sets and the frame-level speech features can be iteratively optimized in the iterative deep learning framework. We call this framework the Multi-granular Acoustic Tokenizing Deep Neural Network (MAT-DNN). The results were evaluated using the metrics and corpora defined in the Zero Resource Speech Challenge organized at Interspeech 2015, and improved performance was obtained with a set of experiments of query-by-example spoken term detection on the same corpora. Visualization for the discovered tokens against the English phonemes was also shown.

No abstract

No abstract