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

Triple M: A Practical Neural Text-to-speech System With Multi-guidance Attention And Multi-band Multi-time Lpcnet

Jan 30, 2021
Shilun Lin, Xinhui Li, Li Lu

Although the sequence-to-sequence network with attention mechanism and neural vocoder has made great progress in the quality of speech synthesis, there are still some problems to be solved in large-scale real-time applications. For example, to avoid long sentence alignment failure while maintaining rich prosody, and to reduce the computational overhead while ensuring perceptual quality. In order to address these issues, we propose a practical neural text-to-speech system, named Triple M, consisting of a seq2seq model with multi-guidance attention and a multi-band multi-time LPCNet. The former uses alignment results of different attention mechanisms to guide the learning of the basic attention mechanism, and only retains the basic attention mechanism during inference. This approach can improve the performance of the text-to-feature module by absorbing the advantages of all guidance attention methods without modifying the basic inference architecture. The latter reduces the computational complexity of LPCNet through combining multi-band and multi-time strategies. The multi-band strategy enables the LPCNet to generate sub-band signals in each inference. By predicting the sub-band signals of adjacent time in one forward operation, the multi-time strategy further decreases the number of inferences required. Due to the multi-band and multi-time strategy, the vocoder speed is increased by 2.75x on a single CPU and the MOS (mean opinion score) degradation is slight.

  Access Paper or Ask Questions

Constructing interval variables via faceted Rasch measurement and multitask deep learning: a hate speech application

Sep 22, 2020
Chris J. Kennedy, Geoff Bacon, Alexander Sahn, Claudia von Vacano

We propose a general method for measuring complex variables on a continuous, interval spectrum by combining supervised deep learning with the Constructing Measures approach to faceted Rasch item response theory (IRT). We decompose the target construct, hate speech in our case, into multiple constituent components that are labeled as ordinal survey items. Those survey responses are transformed via IRT into a debiased, continuous outcome measure. Our method estimates the survey interpretation bias of the human labelers and eliminates that influence on the generated continuous measure. We further estimate the response quality of each labeler using faceted IRT, allowing responses from low-quality labelers to be removed. Our faceted Rasch scaling procedure integrates naturally with a multitask deep learning architecture for automated prediction on new data. The ratings on the theorized components of the target outcome are used as supervised, ordinal variables for the neural networks' internal concept learning. We test the use of an activation function (ordinal softmax) and loss function (ordinal cross-entropy) designed to exploit the structure of ordinal outcome variables. Our multitask architecture leads to a new form of model interpretation because each continuous prediction can be directly explained by the constituent components in the penultimate layer. We demonstrate this new method on a dataset of 50,000 social media comments sourced from YouTube, Twitter, and Reddit and labeled by 11,000 U.S.-based Amazon Mechanical Turk workers to measure a continuous spectrum from hate speech to counterspeech. We evaluate Universal Sentence Encoders, BERT, and RoBERTa as language representation models for the comment text, and compare our predictive accuracy to Google Jigsaw's Perspective API models, showing significant improvement over this standard benchmark.

* 35 pages, 10 figures 

  Access Paper or Ask Questions

Task-specific Optimization of Virtual Channel Linear Prediction-based Speech Dereverberation Front-End for Far-Field Speaker Verification

Dec 27, 2021
Joon-Young Yang, Joon-Hyuk Chang

Developing a single-microphone speech denoising or dereverberation front-end for robust automatic speaker verification (ASV) in noisy far-field speaking scenarios is challenging. To address this problem, we present a novel front-end design that involves a recently proposed extension of the weighted prediction error (WPE) speech dereverberation algorithm, the virtual acoustic channel expansion (VACE)-WPE. It is demonstrated experimentally in this study that unlike the conventional WPE algorithm, the VACE-WPE can be explicitly trained to cancel out both late reverberation and background noise. To build the front-end, the VACE-WPE is first independently (pre)trained to produce "noisy" dereverberated signals. Subsequently, given a pretrained speaker embedding model, the VACE-WPE is additionally fine-tuned within a task-specific optimization (TSO) framework, causing the speaker embedding extracted from the processed signal to be similar to that extracted from the "noise-free" target signal. Moreover, to extend the application of the proposed front-end to more general, unconstrained "in-the-wild" ASV scenarios beyond controlled far-field conditions, we propose a distortion regularization method for the VACE-WPE within the TSO framework. The effectiveness of the proposed approach is verified on both far-field and in-the-wild ASV benchmarks, demonstrating its superiority over fully neural front-ends and other TSO methods in various cases.

  Access Paper or Ask Questions

Data and knowledge-driven approaches for multilingual training to improve the performance of speech recognition systems of Indian languages

Jan 24, 2022
A. Madhavaraj, Ramakrishnan Angarai Ganesan

We propose data and knowledge-driven approaches for multilingual training of the automated speech recognition (ASR) system for a target language by pooling speech data from multiple source languages. Exploiting the acoustic similarities between Indian languages, we implement two approaches. In phone/senone mapping, deep neural network (DNN) learns to map senones or phones from one language to the others, and the transcriptions of the source languages are modified such that they can be used along with the target language data to train and fine-tune the target language ASR system. In the other approach, we model the acoustic information for all the languages simultaneously by training a multitask DNN (MTDNN) to predict the senones of each language in different output layers. The cross-entropy loss and the weight update procedure are modified such that only the shared layers and the output layer responsible for predicting the senone classes of a language are updated during training, if the feature vector belongs to that particular language. In the low-resource setting (LRS), 40 hours of transcribed data each for Tamil, Telugu and Gujarati languages are used for training. The DNN based senone mapping technique gives relative improvements in word error rates (WER) of 9.66%, 7.2% and 15.21% over the baseline system for Tamil, Gujarati and Telugu languages, respectively. In medium-resourced setting (MRS), 160, 275 and 135 hours of data for Tamil, Kannada and Hindi languages are used, where, the same technique gives better relative improvements of 13.94%, 10.28% and 27.24% for Tamil, Kannada and Hindi, respectively. The MTDNN with senone mapping based training in LRS, gives higher relative WER improvements of 15.0%, 17.54% and 16.06%, respectively for Tamil, Gujarati and Telugu, whereas in MRS, we see improvements of 21.24% 21.05% and 30.17% for Tamil, Kannada and Hindi languages, respectively.

  Access Paper or Ask Questions

Bootstrapping Method for Developing Part-of-Speech Tagged Corpus in Low Resource Languages Tagset - A Focus on an African Igbo

Mar 12, 2019
Onyenwe Ikechukwu E, Onyedinma Ebele G, Aniegwu Godwin E, Ezeani Ignatius M

Most languages, especially in Africa, have fewer or no established part-of-speech (POS) tagged corpus. However, POS tagged corpus is essential for natural language processing (NLP) to support advanced researches such as machine translation, speech recognition, etc. Even in cases where there is no POS tagged corpus, there are some languages for which parallel texts are available online. The task of POS tagging a new language corpus with a new tagset usually face a bootstrapping problem at the initial stages of the annotation process. The unavailability of automatic taggers to help the human annotator makes the annotation process to appear infeasible to quickly produce adequate amounts of POS tagged corpus for advanced NLP research and training the taggers. In this paper, we demonstrate the efficacy of a POS annotation method that employed the services of two automatic approaches to assist POS tagged corpus creation for a novel language in NLP. The two approaches are cross-lingual and monolingual POS tags projection. We used cross-lingual to automatically create an initial 'errorful' tagged corpus for a target language via word-alignment. The resources for creating this are derived from a source language rich in NLP resources. A monolingual method is applied to clean the induce noise via an alignment process and to transform the source language tags to the target language tags. We used English and Igbo as our case study. This is possible because there are parallel texts that exist between English and Igbo, and the source language English has available NLP resources. The results of the experiment show a steady improvement in accuracy and rate of tags transformation with score ranges of 6.13% to 83.79% and 8.67% to 98.37% respectively. The rate of tags transformation evaluates the rate at which source language tags are translated to target language tags.

* International Journal on Natural Language Computing (IJNLC) Vol 8(1) (2019) 

  Access Paper or Ask Questions

PM-MMUT: Boosted Phone-mask Data Augmentation using Multi-modeing Unit Training for Robust Uyghur E2E Speech Recognition

Dec 13, 2021
Guodong Ma, Pengfei Hu, Nurmemet Yolwas, Shen Huang, Hao Huang

Consonant and vowel reduction are often encountered in Uyghur speech, which might cause performance degradation in Uyghur automatic speech recognition (ASR). Our recently proposed learning strategy based on masking, Phone Masking Training (PMT), alleviates the impact of such phenomenon in Uyghur ASR. Although PMT achieves remarkably improvements, there still exists room for further gains due to the granularity mismatch between masking unit of PMT (phoneme) and modeling unit (word-piece). To boost the performance of PMT, we propose multi-modeling unit training (MMUT) architecture fusion with PMT (PM-MMUT). The idea of MMUT framework is to split the Encoder into two parts including acoustic feature sequences to phoneme-level representation (AF-to-PLR) and phoneme-level representation to word-piece-level representation (PLR-to-WPLR). It allows AF-to-PLR to be optimized by an intermediate phoneme-based CTC loss to learn the rich phoneme-level context information brought by PMT. Experi-mental results on Uyghur ASR show that the proposed approaches improve significantly, outperforming the pure PMT (reduction WER from 24.0 to 23.7 on Read-Test and from 38.4 to 36.8 on Oral-Test respectively). We also conduct experiments on the 960-hour Librispeech benchmark using ESPnet1, which achieves about 10% relative WER reduction on all the test sets without LM fusion comparing with the latest official ESPnet1 pre-trained model.

* Subbmitted to ICASSP 2022 

  Access Paper or Ask Questions

[email protected]: Using Machine Learning for Detection of Hate Speech and Offensive Code-Mixed Social Media text

Feb 19, 2021
Varsha Pathak, Manish Joshi, Prasad Joshi, Monica Mundada, Tanmay Joshi

This paper describes the system submitted by our team, KBCNMUJAL, for Task 2 of the shared task Hate Speech and Offensive Content Identification in Indo-European Languages (HASOC), at Forum for Information Retrieval Evaluation, December 16-20, 2020, Hyderabad, India. The datasets of two Dravidian languages Viz. Malayalam and Tamil of size 4000 observations, each were shared by the HASOC organizers. These datasets are used to train the machine using different machine learning algorithms, based on classification and regression models. The datasets consist of tweets or YouTube comments with two class labels offensive and not offensive. The machine is trained to classify such social media messages in these two categories. Appropriate n-gram feature sets are extracted to learn the specific characteristics of the Hate Speech text messages. These feature models are based on TFIDF weights of n-gram. The referred work and respective experiments show that the features such as word, character and combined model of word and character n-grams could be used to identify the term patterns of offensive text contents. As a part of the HASOC shared task, the test data sets are made available by the HASOC track organizers. The best performing classification models developed for both languages are applied on test datasets. The model which gives the highest accuracy result on training dataset for Malayalam language was experimented to predict the categories of respective test data. This system has obtained an F1 score of 0.77. Similarly the best performing model for Tamil language has obtained an F1 score of 0.87. This work has received 2nd and 3rd rank in this shared Task 2 for Malayalam and Tamil language respectively. The proposed system is named HASOC_kbcnmujal.

  Access Paper or Ask Questions

GLR-Parsing of Word Lattices Using a Beam Search Method

Jun 22, 1995
Steffen Staab

This paper presents an approach that allows the efficient integration of speech recognition and language understanding using Tomita's generalized LR-parsing algorithm. For this purpose the GLRP-algorithm is revised so that an agenda mechanism can be used to control the flow of computation of the parsing process. This new approach is used to integrate speech recognition and speech understanding incrementally with a beam search method. These considerations have been implemented and tested on ten word lattices.

* 4 pages, 61K postscript, compressed, uuencoded, Eurospeech 9/95, Madrid 

  Access Paper or Ask Questions

Get out the vote: Determining support or opposition from Congressional floor-debate transcripts

Jun 06, 2012
Matt Thomas, Bo Pang, Lillian Lee

We investigate whether one can determine from the transcripts of U.S. Congressional floor debates whether the speeches represent support of or opposition to proposed legislation. To address this problem, we exploit the fact that these speeches occur as part of a discussion; this allows us to use sources of information regarding relationships between discourse segments, such as whether a given utterance indicates agreement with the opinion expressed by another. We find that the incorporation of such information yields substantial improvements over classifying speeches in isolation.

* Proceedings of EMNLP 2006. This revision (created Dec 2006) contains clearly-marked updates from the proceedings version. Data available at 

  Access Paper or Ask Questions

Honk: A PyTorch Reimplementation of Convolutional Neural Networks for Keyword Spotting

Nov 28, 2017
Raphael Tang, Jimmy Lin

We describe Honk, an open-source PyTorch reimplementation of convolutional neural networks for keyword spotting that are included as examples in TensorFlow. These models are useful for recognizing "command triggers" in speech-based interfaces (e.g., "Hey Siri"), which serve as explicit cues for audio recordings of utterances that are sent to the cloud for full speech recognition. Evaluation on Google's recently released Speech Commands Dataset shows that our reimplementation is comparable in accuracy and provides a starting point for future work on the keyword spotting task.

* 3 pages, 2 figures 

  Access Paper or Ask Questions