Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox

"speech": models, code, and papers

Fast, Compact, and High Quality LSTM-RNN Based Statistical Parametric Speech Synthesizers for Mobile Devices

Jun 22, 2016
Heiga Zen, Yannis Agiomyrgiannakis, Niels Egberts, Fergus Henderson, Przemysław Szczepaniak

Acoustic models based on long short-term memory recurrent neural networks (LSTM-RNNs) were applied to statistical parametric speech synthesis (SPSS) and showed significant improvements in naturalness and latency over those based on hidden Markov models (HMMs). This paper describes further optimizations of LSTM-RNN-based SPSS for deployment on mobile devices; weight quantization, multi-frame inference, and robust inference using an {\epsilon}-contaminated Gaussian loss function. Experimental results in subjective listening tests show that these optimizations can make LSTM-RNN-based SPSS comparable to HMM-based SPSS in runtime speed while maintaining naturalness. Evaluations between LSTM-RNN- based SPSS and HMM-driven unit selection speech synthesis are also presented.

* 13 pages, 3 figures, Interspeech 2016 (accepted) 

  Access Paper or Ask Questions

Non-linear frequency warping using constant-Q transformation for speech emotion recognition

Feb 08, 2021
Premjeet Singh, Goutam Saha, Md Sahidullah

In this work, we explore the constant-Q transform (CQT) for speech emotion recognition (SER). The CQT-based time-frequency analysis provides variable spectro-temporal resolution with higher frequency resolution at lower frequencies. Since lower-frequency regions of speech signal contain more emotion-related information than higher-frequency regions, the increased low-frequency resolution of CQT makes it more promising for SER than standard short-time Fourier transform (STFT). We present a comparative analysis of short-term acoustic features based on STFT and CQT for SER with deep neural network (DNN) as a back-end classifier. We optimize different parameters for both features. The CQT-based features outperform the STFT-based spectral features for SER experiments. Further experiments with cross-corpora evaluation demonstrate that the CQT-based systems provide better generalization with out-of-domain training data.

* Accepted for publication in 2021 IEEE International Conference on Computer Communication and Informatics (IEEE ICCCI 2021) 

  Access Paper or Ask Questions

Contextual Biasing of Language Models for Speech Recognition in Goal-Oriented Conversational Agents

Mar 19, 2021
Ashish Shenoy, Sravan Bodapati, Katrin Kirchhoff

Goal-oriented conversational interfaces are designed to accomplish specific tasks and typically have interactions that tend to span multiple turns adhering to a pre-defined structure and a goal. However, conventional neural language models (NLM) in Automatic Speech Recognition (ASR) systems are mostly trained sentence-wise with limited context. In this paper, we explore different ways to incorporate context into a LSTM based NLM in order to model long range dependencies and improve speech recognition. Specifically, we use context carry over across multiple turns and use lexical contextual cues such as system dialog act from Natural Language Understanding (NLU) models and the user provided structure of the chatbot. We also propose a new architecture that utilizes context embeddings derived from BERT on sample utterances provided during inference time. Our experiments show a word error rate (WER) relative reduction of 7% over non-contextual utterance-level NLM rescorers on goal-oriented audio datasets.

* 7 pages 

  Access Paper or Ask Questions

Constructing Long Short-Term Memory based Deep Recurrent Neural Networks for Large Vocabulary Speech Recognition

May 11, 2015
Xiangang Li, Xihong Wu

Long short-term memory (LSTM) based acoustic modeling methods have recently been shown to give state-of-the-art performance on some speech recognition tasks. To achieve a further performance improvement, in this research, deep extensions on LSTM are investigated considering that deep hierarchical model has turned out to be more efficient than a shallow one. Motivated by previous research on constructing deep recurrent neural networks (RNNs), alternative deep LSTM architectures are proposed and empirically evaluated on a large vocabulary conversational telephone speech recognition task. Meanwhile, regarding to multi-GPU devices, the training process for LSTM networks is introduced and discussed. Experimental results demonstrate that the deep LSTM networks benefit from the depth and yield the state-of-the-art performance on this task.

* submitted to ICASSP 2015 which does not perform blind reviews 

  Access Paper or Ask Questions

NaturalSpeech: End-to-End Text to Speech Synthesis with Human-Level Quality

May 10, 2022
Xu Tan, Jiawei Chen, Haohe Liu, Jian Cong, Chen Zhang, Yanqing Liu, Xi Wang, Yichong Leng, Yuanhao Yi, Lei He, Frank Soong, Tao Qin, Sheng Zhao, Tie-Yan Liu

Text to speech (TTS) has made rapid progress in both academia and industry in recent years. Some questions naturally arise that whether a TTS system can achieve human-level quality, how to define/judge that quality and how to achieve it. In this paper, we answer these questions by first defining the human-level quality based on the statistical significance of subjective measure and introducing appropriate guidelines to judge it, and then developing a TTS system called NaturalSpeech that achieves human-level quality on a benchmark dataset. Specifically, we leverage a variational autoencoder (VAE) for end-to-end text to waveform generation, with several key modules to enhance the capacity of the prior from text and reduce the complexity of the posterior from speech, including phoneme pre-training, differentiable duration modeling, bidirectional prior/posterior modeling, and a memory mechanism in VAE. Experiment evaluations on popular LJSpeech dataset show that our proposed NaturalSpeech achieves -0.01 CMOS (comparative mean opinion score) to human recordings at the sentence level, with Wilcoxon signed rank test at p-level p >> 0.05, which demonstrates no statistically significant difference from human recordings for the first time on this dataset.

* 19 pages, 3 figures, 8 tables 

  Access Paper or Ask Questions

Revisiting the Application of Feature Selection Methods to Speech Imagery BCI Datasets

Aug 17, 2020
Javad Rahimipour Anaraki, Jae Moon, Tom Chau

Brain-computer interface (BCI) aims to establish and improve human and computer interactions. There has been an increasing interest in designing new hardware devices to facilitate the collection of brain signals through various technologies, such as wet and dry electroencephalogram (EEG) and functional near-infrared spectroscopy (fNIRS) devices. The promising results of machine learning methods have attracted researchers to apply these methods to their data. However, some methods can be overlooked simply due to their inferior performance against a particular dataset. This paper shows how relatively simple yet powerful feature selection/ranking methods can be applied to speech imagery datasets and generate significant results. To do so, we introduce two approaches, horizontal and vertical settings, to use any feature selection and ranking methods to speech imagery BCI datasets. Our primary goal is to improve the resulting classification accuracies from support vector machines, $k$-nearest neighbour, decision tree, linear discriminant analysis and long short-term memory recurrent neural network classifiers. Our experimental results show that using a small subset of channels, we can retain and, in most cases, improve the resulting classification accuracies regardless of the classifier.

* 5 pages, 2 figures 

  Access Paper or Ask Questions

Detecting Hate Speech in Memes Using Multimodal Deep Learning Approaches: Prize-winning solution to Hateful Memes Challenge

Dec 23, 2020
Riza Velioglu, Jewgeni Rose

Memes on the Internet are often harmless and sometimes amusing. However, by using certain types of images, text, or combinations of both, the seemingly harmless meme becomes a multimodal type of hate speech -- a hateful meme. The Hateful Memes Challenge is a first-of-its-kind competition which focuses on detecting hate speech in multimodal memes and it proposes a new data set containing 10,000+ new examples of multimodal content. We utilize VisualBERT -- which meant to be the BERT of vision and language -- that was trained multimodally on images and captions and apply Ensemble Learning. Our approach achieves 0.811 AUROC with an accuracy of 0.765 on the challenge test set and placed third out of 3,173 participants in the Hateful Memes Challenge.

* Presented at NeurIPS (2020) 

  Access Paper or Ask Questions

A comparison of streaming models and data augmentation methods for robust speech recognition

Nov 19, 2021
Jiyeon Kim, Mehul Kumar, Dhananjaya Gowda, Abhinav Garg, Chanwoo Kim

In this paper, we present a comparative study on the robustness of two different online streaming speech recognition models: Monotonic Chunkwise Attention (MoChA) and Recurrent Neural Network-Transducer (RNN-T). We explore three recently proposed data augmentation techniques, namely, multi-conditioned training using an acoustic simulator, Vocal Tract Length Perturbation (VTLP) for speaker variability, and SpecAugment. Experimental results show that unidirectional models are in general more sensitive to noisy examples in the training set. It is observed that the final performance of the model depends on the proportion of training examples processed by data augmentation techniques. MoChA models generally perform better than RNN-T models. However, we observe that training of MoChA models seems to be more sensitive to various factors such as the characteristics of training sets and the incorporation of additional augmentations techniques. On the other hand, RNN-T models perform better than MoChA models in terms of latency, inference time, and the stability of training. Additionally, RNN-T models are generally more robust against noise and reverberation. All these advantages make RNN-T models a better choice for streaming on-device speech recognition compared to MoChA models.

* Accepted as a conference paper at ASRU 2021 

  Access Paper or Ask Questions

Simplified Self-Attention for Transformer-based End-to-End Speech Recognition

May 21, 2020
Haoneng Luo, Shiliang Zhang, Ming Lei, Lei Xie

Transformer models have been introduced into end-to-end speech recognition with state-of-the-art performance on various tasks owing to their superiority in modeling long-term dependencies. However, such improvements are usually obtained through the use of very large neural networks. Transformer models mainly include two submodules - position-wise feedforward layers and self-attention (SAN) layers. In this paper, to reduce the model complexity while maintaining good performance, we propose a simplified self-attention (SSAN) layer which employs FSMN memory block instead of projection layers to form query and key vectors for transformer-based end-to-end speech recognition. We evaluate the SSAN-based and the conventional SAN-based transformers on the public AISHELL-1, internal 1000-hour and 20,000-hour large-scale Mandarin tasks. Results show that our proposed SSAN-based transformer model can achieve over 20% relative reduction in model parameters and 6.7% relative CER reduction on the AISHELL-1 task. With impressively 20% parameter reduction, our model shows no loss of recognition performance on the 20,000-hour large-scale task.

* Submitted to Interspeech2020 

  Access Paper or Ask Questions