arXiv subjects

Jaejin Cho

Publications and source records attributed to Jaejin Cho.

25 records · Page 2Linked to original sources

arXiv2020

Learning Speaker Embedding from Text-to-Speech

Zero-shot multi-speaker Text-to-Speech (TTS) generates target speaker voices given an input text and the corresponding speaker embedding. In this work, we investigate the effectiveness of the TTS reconstruction objective to improve representation learning for speaker verification. We jointly trained end-to-end Tacotron 2 TTS and speaker embedding networks in a self-supervised fashion. We hypothesize that the embeddings will contain minimal phonetic information since the TTS decoder will obtain that information from the textual input. TTS reconstruction can also be combined with speaker classification to enhance these embeddings further. Once trained, the speaker encoder computes representations for the speaker verification task, while the rest of the TTS blocks are discarded. We investigated training TTS from either manual or ASR-generated transcripts. The latter allows us to train embeddings on datasets without manual transcripts. We compared ASR transcripts and Kaldi phone alignments as TTS inputs, showing that the latter performed better due to their finer resolution. Unsupervised TTS embeddings improved EER by 2.06\% absolute with regard to i-vectors for the LibriTTS dataset. TTS with speaker classification loss improved EER by 0.28\% and 0.73\% absolutely from a model using only speaker classification loss in LibriTTS and Voxceleb1 respectively.

eess.AS↗

arXiv2019

Scan-specific, Parameter-free Artifact Reduction in K-space (SPARK)

We propose a convolutional neural network (CNN) approach that works synergistically with physics-based reconstruction methods to reduce artifacts in accelerated MRI. Given reconstructed coil k-spaces, our network predicts a k-space correction term for each coil. This is done by matching the difference between the acquired autocalibration lines and their erroneous reconstructions, and generalizing this error term over the entire k-space. Application of this approach on existing reconstruction methods show that SPARK suppresses reconstruction artifacts at high acceleration, while preserving and improving on detail in moderate acceleration rates where existing reconstruction algorithms already perform well; indicating robustness. Introduction Parallel

eess.IV↗

arXiv2019

Deep neural networks for emotion recognition combining audio and transcripts

In this paper, we propose to improve emotion recognition by combining acoustic information and conversation transcripts. On the one hand, an LSTM network was used to detect emotion from acoustic features like f0, shimmer, jitter, MFCC, etc. On the other hand, a multi-resolution CNN was used to detect emotion from word sequences. This CNN consists of several parallel convolutions with different kernel sizes to exploit contextual information at different levels. A temporal pooling layer aggregates the hidden representations of different words into a unique sequence level embedding, from which we computed the emotion posteriors. We optimized a weighted sum of classification and verification losses. The verification loss tries to bring embeddings from the same emotions closer while separating embeddings from different emotions. We also compared our CNN with state-of-the-art text-based hand-crafted features (e-vector). We evaluated our approach on the USC-IEMOCAP dataset as well as the dataset consisting of US English telephone speech. In the former, we used human-annotated transcripts while in the latter, we used ASR transcripts. The results showed fusing audio and transcript information improved unweighted accuracy by relative 24% for IEMOCAP and relative 3.4% for the telephone data compared to a single acoustic system.

eess.AS↗

arXiv2019

Highly efficient MRI through multi-shot echo planar imaging

Multi-shot echo planar imaging (msEPI) is a promising approach to achieve high in-plane resolution with high sampling efficiency and low T2* blurring. However, due to the geometric distortion, shot-to-shot phase variations and potential subject motion, msEPI continues to be a challenge in MRI. In this work, we introduce acquisition and reconstruction strategies for robust, high-quality msEPI without phase navigators. We propose Blip Up-Down Acquisition (BUDA) using interleaved blip-up and -down phase encoding, and incorporate B0 forward-modeling into Hankel structured low-rank model to enable distortion- and navigator-free msEPI. We improve the acquisition efficiency and reconstruction quality by incorporating simultaneous multi-slice acquisition and virtual-coil reconstruction into the BUDA technique. We further combine BUDA with the novel RF-encoded gSlider acquisition, dubbed BUDA-gSlider, to achieve rapid high isotropic-resolution MRI. Deploying BUDA-gSlider with model-based reconstruction allows for distortion-free whole-brain 1mm isotropic T2 mapping in about 1 minute. It also provides whole-brain 1mm isotropic diffusion imaging with high geometric fidelity and SNR efficiency. We finally incorporate sinusoidal wave gradients during the EPI readout to better use coil sensitivity encoding with controlled aliasing.

eess.IV↗

arXiv2019

Transfer learning of language-independent end-to-end ASR with language model fusion

This work explores better adaptation methods to low-resource languages using an external language model (LM) under the framework of transfer learning. We first build a language-independent ASR system in a unified sequence-to-sequence (S2S) architecture with a shared vocabulary among all languages. During adaptation, we perform LM fusion transfer, where an external LM is integrated into the decoder network of the attention-based S2S model in the whole adaptation stage, to effectively incorporate linguistic context of the target language. We also investigate various seed models for transfer learning. Experimental evaluations using the IARPA BABEL data set show that LM fusion transfer improves performances on all target five languages compared with simple transfer learning when the external text data is available. Our final system drastically reduces the performance gap from the hybrid systems.

cs.CL↗

arXiv2018

Multilingual sequence-to-sequence speech recognition: architecture, transfer learning, and language modeling

Sequence-to-sequence (seq2seq) approach for low-resource ASR is a relatively new direction in speech research. The approach benefits by performing model training without using lexicon and alignments. However, this poses a new problem of requiring more data compared to conventional DNN-HMM systems. In this work, we attempt to use data from 10 BABEL languages to build a multi-lingual seq2seq model as a prior model, and then port them towards 4 other BABEL languages using transfer learning approach. We also explore different architectures for improving the prior multilingual seq2seq model. The paper also discusses the effect of integrating a recurrent neural network language model (RNNLM) with a seq2seq model during decoding. Experimental results show that the transfer learning approach from the multilingual model shows substantial gains over monolingual models across all 4 BABEL languages. Incorporating an RNNLM also brings significant improvements in terms of %WER, and achieves recognition performance comparable to the models trained with twice more training data.

cs.CL↗

arXiv2018

End-to-end Speech Recognition with Word-based RNN Language Models

This paper investigates the impact of word-based RNN language models (RNN-LMs) on the performance of end-to-end automatic speech recognition (ASR). In our prior work, we have proposed a multi-level LM, in which character-based and word-based RNN-LMs are combined in hybrid CTC/attention-based ASR. Although this multi-level approach achieves significant error reduction in the Wall Street Journal (WSJ) task, two different LMs need to be trained and used for decoding, which increase the computational cost and memory usage. In this paper, we further propose a novel word-based RNN-LM, which allows us to decode with only the word-based LM, where it provides look-ahead word probabilities to predict next characters instead of the character-based LM, leading competitive accuracy with less computation compared to the multi-level LM. We demonstrate the efficacy of the word-based RNN-LMs using a larger corpus, LibriSpeech, in addition to WSJ we used in the prior work. Furthermore, we show that the proposed model achieves 5.1 %WER for WSJ Eval'92 test set when the vocabulary size is increased, which is the best WER reported for end-to-end ASR systems on this benchmark.

cs.CL↗