Bilingual English speakers speak English as one of their languages. Their English is of a non-native kind, and their conversations are of a code-mixed fashion. The intelligibility of a bilingual text-to-speech (TTS) system for such non-native English speakers depends on a lexicon that captures the phoneme sequence used by non-native speakers. However, due to the lack of non-native English lexicon, existing bilingual TTS systems employ native English lexicons that are widely available, in addition to their native language lexicon. Due to the inconsistency between the non-native English pronunciation in the audio and native English lexicon in the text, the intelligibility of synthesized speech in such TTS systems is significantly reduced. This paper is motivated by the knowledge that the native language of the speaker highly influences non-native English pronunciation. We propose a generic approach to obtain rules based on letter to phoneme alignment to map native English lexicon to their non-native version. The effectiveness of such mapping is studied by comparing bilingual (Indian English and Hindi) TTS systems trained with and without the proposed rules. The subjective evaluation shows that the bilingual TTS system trained with the proposed non-native English lexicon rules obtains a 6% absolute improvement in preference.
Articulatory-to-acoustic (forward) mapping is a technique to predict speech using various articulatory acquisition techniques as input (e.g. ultrasound tongue imaging, MRI, lip video). The advantage of lip video is that it is easily available and affordable: most modern smartphones have a front camera. There are already a few solutions for lip-to-speech synthesis, but they mostly concentrate on offline training and inference. In this paper, we propose a system built from a backend for deep neural network training and inference and a fronted as a form of a mobile application. Our initial evaluation shows that the scenario is feasible: a top-5 classification accuracy of 74% is combined with feedback from the mobile application user, making sure that the speaking impaired might be able to communicate with this solution.
Automatic speech recognition systems have accomplished remarkable improvements in transcription accuracy in recent years. On some domains, models now achieve near-human performance. However, transcription performance on oral history has not yet reached human accuracy. In the present work, we investigate how large this gap between human and machine transcription still is. For this purpose, we analyze and compare transcriptions of three humans on a new oral history data set. We estimate a human word error rate of 8.7% for recent German oral history interviews with clean acoustic conditions. For comparison with recent machine transcription accuracy, we present experiments on the adaptation of an acoustic model achieving near-human performance on broadcast speech. We investigate the influence of different adaptation data on robustness and generalization for clean and noisy oral history interviews. We optimize our acoustic models by 5 to 8% relative for this task and achieve 23.9% WER on noisy and 15.6% word error rate on clean oral history interviews.
In this work, we focus on improving ASR output segmentation in the context of low-resource language speech-to-text translation. ASR output segmentation is crucial, as ASR systems segment the input audio using purely acoustic information and are not guaranteed to output sentence-like segments. Since most MT systems expect sentences as input, feeding in longer unsegmented passages can lead to sub-optimal performance. We explore the feasibility of using datasets of subtitles from TV shows and movies to train better ASR segmentation models. We further incorporate part-of-speech (POS) tag and dependency label information (derived from the unsegmented ASR outputs) into our segmentation model. We show that this noisy syntactic information can improve model accuracy. We evaluate our models intrinsically on segmentation quality and extrinsically on downstream MT performance, as well as downstream tasks including cross-lingual information retrieval (CLIR) tasks and human relevance assessments. Our model shows improved performance on downstream tasks for Lithuanian and Bulgarian.
In this work, we propose "global style tokens" (GSTs), a bank of embeddings that are jointly trained within Tacotron, a state-of-the-art end-to-end speech synthesis system. The embeddings are trained with no explicit labels, yet learn to model a large range of acoustic expressiveness. GSTs lead to a rich set of significant results. The soft interpretable "labels" they generate can be used to control synthesis in novel ways, such as varying speed and speaking style - independently of the text content. They can also be used for style transfer, replicating the speaking style of a single audio clip across an entire long-form text corpus. When trained on noisy, unlabeled found data, GSTs learn to factorize noise and speaker identity, providing a path towards highly scalable but robust speech synthesis.
In this work, we present a database of multimodal communication features extracted from debate speeches in the 2019 North American Universities Debate Championships (NAUDC). Feature sets were extracted from the visual (facial expression, gaze, and head pose), audio (PRAAT), and textual (word sentiment and linguistic category) modalities of raw video recordings of competitive collegiate debaters (N=717 6-minute recordings from 140 unique debaters). Each speech has an associated competition debate score (range: 67-96) from expert judges as well as competitor demographic and per-round reflection surveys. We observe the fully multimodal model performs best in comparison to models trained on various compositions of modalities. We also find that the weights of some features (such as the expression of joy and the use of the word we) change in direction between the aforementioned models. We use these results to highlight the value of a multimodal dataset for studying competitive, collegiate debate.
We present a neural analysis and synthesis (NANSY) framework that can manipulate voice, pitch, and speed of an arbitrary speech signal. Most of the previous works have focused on using information bottleneck to disentangle analysis features for controllable synthesis, which usually results in poor reconstruction quality. We address this issue by proposing a novel training strategy based on information perturbation. The idea is to perturb information in the original input signal (e.g., formant, pitch, and frequency response), thereby letting synthesis networks selectively take essential attributes to reconstruct the input signal. Because NANSY does not need any bottleneck structures, it enjoys both high reconstruction quality and controllability. Furthermore, NANSY does not require any labels associated with speech data such as text and speaker information, but rather uses a new set of analysis features, i.e., wav2vec feature and newly proposed pitch feature, Yingram, which allows for fully self-supervised training. Taking advantage of fully self-supervised training, NANSY can be easily extended to a multilingual setting by simply training it with a multilingual dataset. The experiments show that NANSY can achieve significant improvement in performance in several applications such as zero-shot voice conversion, pitch shift, and time-scale modification.
Continual learning (CL), or domain expansion, recently became a popular topic for automatic speech recognition (ASR) acoustic modeling because practical systems have to be updated frequently in order to work robustly on types of speech not observed during initial training. While sequential adaptation allows tuning a system to a new domain, it may result in performance degradation on the old domains due to catastrophic forgetting. In this work we explore regularization-based CL for neural network acoustic models trained with the lattice-free maximum mutual information (LF-MMI) criterion. We simulate domain expansion by incrementally adapting the acoustic model on different public datasets that include several accents and speaking styles. We investigate two well-known CL techniques, elastic weight consolidation (EWC) and learning without forgetting (LWF), which aim to reduce forgetting by preserving model weights or network outputs. We additionally introduce a sequence-level LWF regularization, which exploits posteriors from the denominator graph of LF-MMI to further reduce forgetting. Empirical results show that the proposed sequence-level LWF can improve the best average word error rate across all domains by up to 9.4% relative compared with using regular LWF.
Self-attention models have been successfully applied in end-to-end speech recognition systems, which greatly improve the performance of recognition accuracy. However, such attention-based models cannot be used in online speech recognition, because these models usually have to utilize a whole acoustic sequences as inputs. A common method is restricting the field of attention sights by a fixed left and right window, which makes the computation costs manageable yet also introduces performance degradation. In this paper, we propose Memory-Self-Attention (MSA), which adds history information into the Restricted-Self-Attention unit. MSA only needs localtime features as inputs, and efficiently models long temporal contexts by attending memory states. Meanwhile, recurrent neural network transducer (RNN-T) has proved to be a great approach for online ASR tasks, because the alignments of RNN-T are local and monotonic. We propose a novel network structure, called Memory-Self-Attention (MSA) Transducer. Both encoder and decoder of the MSA Transducer contain the proposed MSA unit. The experiments demonstrate that our proposed models improve WER results than Restricted-Self-Attention models by $13.5 on WSJ and $7.1 on SWBD datasets relatively, and without much computation costs increase.
The Aduio-visual Speech Recognition (AVSR) which employs both the video and audio information to do Automatic Speech Recognition (ASR) is one of the application of multimodal leaning making ASR system more robust and accuracy. The traditional models usually treated AVSR as inference or projection but strict prior limits its ability. As the revival of deep learning, Deep Neural Networks (DNN) becomes an important toolkit in many traditional classification tasks including ASR, image classification, natural language processing. Some DNN models were used in AVSR like Multimodal Deep Autoencoders (MDAEs), Multimodal Deep Belief Network (MDBN) and Multimodal Deep Boltzmann Machine (MDBM) that actually work better than traditional methods. However, such DNN models have several shortcomings: (1) They don't balance the modal fusion and temporal fusion, or even haven't temporal fusion; (2)The architecture of these models isn't end-to-end, the training and testing getting cumbersome. We propose a DNN model, Auxiliary Multimodal LSTM (am-LSTM), to overcome such weakness. The am-LSTM could be trained and tested once, moreover easy to train and preventing overfitting automatically. The extensibility and flexibility are also take into consideration. The experiments show that am-LSTM is much better than traditional methods and other DNN models in three datasets.