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

Non-Autoregressive TTS with Explicit Duration Modelling for Low-Resource Highly Expressive Speech

Jun 24, 2021
Raahil Shah, Kamil Pokora, Abdelhamid Ezzerg, Viacheslav Klimkov, Goeric Huybrechts, Bartosz Putrycz, Daniel Korzekwa, Thomas Merritt

Whilst recent neural text-to-speech (TTS) approaches produce high-quality speech, they typically require a large amount of recordings from the target speaker. In previous work, a 3-step method was proposed to generate high-quality TTS while greatly reducing the amount of data required for training. However, we have observed a ceiling effect in the level of naturalness achievable for highly expressive voices when using this approach. In this paper, we present a method for building highly expressive TTS voices with as little as 15 minutes of speech data from the target speaker. Compared to the current state-of-the-art approach, our proposed improvements close the gap to recordings by 23.3% for naturalness of speech and by 16.3% for speaker similarity. Further, we match the naturalness and speaker similarity of a Tacotron2-based full-data (~10 hours) model using only 15 minutes of target speaker data, whereas with 30 minutes or more, we significantly outperform it. The following improvements are proposed: 1) changing from an autoregressive, attention-based TTS model to a non-autoregressive model replacing attention with an external duration model and 2) an additional Conditional Generative Adversarial Network (cGAN) based fine-tuning step.

* 6 pages, 5 figures. Accepted to Speech Synthesis Workshop (SSW) 2021 

  Access Paper or Ask Questions

Combination of Time-domain, Frequency-domain, and Cepstral-domain Acoustic Features for Speech Commands Classification

Mar 30, 2022
Yikang Wang, Hiromitsu Nishizaki

In speech-related classification tasks, frequency-domain acoustic features such as logarithmic Mel-filter bank coefficients (FBANK) and cepstral-domain acoustic features such as Mel-frequency cepstral coefficients (MFCC) are often used. However, time-domain features perform more effectively in some sound classification tasks which contain non-vocal or weakly speech-related sounds. We previously proposed a feature called bit sequence representation (BSR), which is a time-domain binary acoustic feature based on the raw waveform. Compared with MFCC, BSR performed better in environmental sound detection and showed comparable accuracy performance in limited-vocabulary speech recognition tasks. In this paper, we propose a novel improvement BSR feature called BSR-float16 to represent floating-point values more precisely. We experimentally demonstrated the complementarity among time-domain, frequency-domain, and cepstral-domain features using a dataset called Speech Commands proposed by Google. Therefore, we used a simple back-end score fusion method to improve the final classification accuracy. The fusion results also showed better noise robustness.

* 5 pages, 4 figures. This paper has been submitted to Interspeech 2022 

  Access Paper or Ask Questions

AIPNet: Generative Adversarial Pre-training of Accent-invariant Networks for End-to-end Speech Recognition

Nov 27, 2019
Yi-Chen Chen, Zhaojun Yang, Ching-Feng Yeh, Mahaveer Jain, Michael L. Seltzer

As one of the major sources in speech variability, accents have posed a grand challenge to the robustness of speech recognition systems. In this paper, our goal is to build a unified end-to-end speech recognition system that generalizes well across accents. For this purpose, we propose a novel pre-training framework AIPNet based on generative adversarial nets (GAN) for accent-invariant representation learning: Accent Invariant Pre-training Networks. We pre-train AIPNet to disentangle accent-invariant and accent-specific characteristics from acoustic features through adversarial training on accented data for which transcriptions are not necessarily available. We further fine-tune AIPNet by connecting the accent-invariant module with an attention-based encoder-decoder model for multi-accent speech recognition. In the experiments, our approach is compared against four baselines including both accent-dependent and accent-independent models. Experimental results on 9 English accents show that the proposed approach outperforms all the baselines by 2.3 \sim 4.5% relative reduction on average WER when transcriptions are available in all accents and by 1.6 \sim 6.1% relative reduction when transcriptions are only available in US accent.

  Access Paper or Ask Questions

Mixture factorized auto-encoder for unsupervised hierarchical deep factorization of speech signal

Oct 30, 2019
Zhiyuan Peng, Siyuan Feng, Tan Lee

Speech signal is constituted and contributed by various informative factors, such as linguistic content and speaker characteristic. There have been notable recent studies attempting to factorize speech signal into these individual factors without requiring any annotation. These studies typically assume continuous representation for linguistic content, which is not in accordance with general linguistic knowledge and may make the extraction of speaker information less successful. This paper proposes the mixture factorized auto-encoder (mFAE) for unsupervised deep factorization. The encoder part of mFAE comprises a frame tokenizer and an utterance embedder. The frame tokenizer models linguistic content of input speech with a discrete categorical distribution. It performs frame clustering by assigning each frame a soft mixture label. The utterance embedder generates an utterance-level vector representation. A frame decoder serves to reconstruct speech features from the encoders'outputs. The mFAE is evaluated on speaker verification (SV) task and unsupervised subword modeling (USM) task. The SV experiments on VoxCeleb 1 show that the utterance embedder is capable of extracting speaker-discriminative embeddings with performance comparable to a x-vector baseline. The USM experiments on ZeroSpeech 2017 dataset verify that the frame tokenizer is able to capture linguistic content and the utterance embedder can acquire speaker-related information.

  Access Paper or Ask Questions

Speech Gesture Generation from the Trimodal Context of Text, Audio, and Speaker Identity

Sep 04, 2020
Youngwoo Yoon, Bok Cha, Joo-Haeng Lee, Minsu Jang, Jaeyeon Lee, Jaehong Kim, Geehyuk Lee

For human-like agents, including virtual avatars and social robots, making proper gestures while speaking is crucial in human--agent interaction. Co-speech gestures enhance interaction experiences and make the agents look alive. However, it is difficult to generate human-like gestures due to the lack of understanding of how people gesture. Data-driven approaches attempt to learn gesticulation skills from human demonstrations, but the ambiguous and individual nature of gestures hinders learning. In this paper, we present an automatic gesture generation model that uses the multimodal context of speech text, audio, and speaker identity to reliably generate gestures. By incorporating a multimodal context and an adversarial training scheme, the proposed model outputs gestures that are human-like and that match with speech content and rhythm. We also introduce a new quantitative evaluation metric for gesture generation models. Experiments with the introduced metric and subjective human evaluation showed that the proposed gesture generation model is better than existing end-to-end generation models. We further confirm that our model is able to work with synthesized audio in a scenario where contexts are constrained, and show that different gesture styles can be generated for the same speech by specifying different speaker identities in the style embedding space that is learned from videos of various speakers. All the code and data is available at

* 16 pages; ACM Transactions on Graphics (SIGGRAPH Asia 2020) 

  Access Paper or Ask Questions

Heterogeneous Separation Consistency Training for Adaptation of Unsupervised Speech Separation

Apr 23, 2022
Jiangyu Han, Yanhua Long

Recently, supervised speech separation has made great progress. However, limited by the nature of supervised training, most existing separation methods require ground-truth sources and are trained on synthetic datasets. This ground-truth reliance is problematic, because the ground-truth signals are usually unavailable in real conditions. Moreover, in many industry scenarios, the real acoustic characteristics deviate far from the ones in simulated datasets. Therefore, the performance usually degrades significantly when applying the supervised speech separation models to real applications. To address these problems, in this study, we propose a novel separation consistency training, termed SCT, to exploit the real-world unlabeled mixtures for improving cross-domain unsupervised speech separation in an iterative manner, by leveraging upon the complementary information obtained from heterogeneous (structurally distinct but behaviorally complementary) models. SCT follows a framework using two heterogeneous neural networks (HNNs) to produce high confidence pseudo labels of unlabeled real speech mixtures. These labels are then updated, and used to refine the HNNs to produce more reliable consistent separation results for real mixture pseudo-labeling. To maximally utilize the large complementary information between different separation networks, a cross-knowledge adaptation is further proposed. Together with simulated dataset, those real mixtures with high confidence pseudo labels are then used to update the HNN separation models iteratively. In addition, we find that combing the heterogeneous separation outputs by a simple linear fusion can further slightly improve the final system performance.

* submitted to TASLP; omit some abstracts 

  Access Paper or Ask Questions

Streaming on-device detection of device directed speech from voice and touch-based invocation

Oct 09, 2021
Ognjen Rudovic, Akanksha Bindal, Vineet Garg, Pramod Simha, Pranay Dighe, Sachin Kajarekar

When interacting with smart devices such as mobile phones or wearables, the user typically invokes a virtual assistant (VA) by saying a keyword or by pressing a button on the device. However, in many cases, the VA can accidentally be invoked by the keyword-like speech or accidental button press, which may have implications on user experience and privacy. To this end, we propose an acoustic false-trigger-mitigation (FTM) approach for on-device device-directed speech detection that simultaneously handles the voice-trigger and touch-based invocation. To facilitate the model deployment on-device, we introduce a new streaming decision layer, derived using the notion of temporal convolutional networks (TCN) [1], known for their computational efficiency. To the best of our knowledge, this is the first approach that can detect device-directed speech from more than one invocation type in a streaming fashion. We compare this approach with streaming alternatives based on vanilla Average layer, and canonical LSTMs, and show: (i) that all the models show only a small degradation in accuracy compared with the invocation-specific models, and (ii) that the newly introduced streaming TCN consistently performs better or comparable with the alternatives, while mitigating device undirected speech faster in time, and with (relative) reduction in runtime peak-memory over the LSTM-based approach of 33% vs. 7%, when compared to a non-streaming counterpart.

  Access Paper or Ask Questions

Generating diverse and natural text-to-speech samples using a quantized fine-grained VAE and auto-regressive prosody prior

Feb 06, 2020
Guangzhi Sun, Yu Zhang, Ron J. Weiss, Yuan Cao, Heiga Zen, Andrew Rosenberg, Bhuvana Ramabhadran, Yonghui Wu

Recent neural text-to-speech (TTS) models with fine-grained latent features enable precise control of the prosody of synthesized speech. Such models typically incorporate a fine-grained variational autoencoder (VAE) structure, extracting latent features at each input token (e.g., phonemes). However, generating samples with the standard VAE prior often results in unnatural and discontinuous speech, with dramatic prosodic variation between tokens. This paper proposes a sequential prior in a discrete latent space which can generate more naturally sounding samples. This is accomplished by discretizing the latent features using vector quantization (VQ), and separately training an autoregressive (AR) prior model over the result. We evaluate the approach using listening tests, objective metrics of automatic speech recognition (ASR) performance, and measurements of prosody attributes. Experimental results show that the proposed model significantly improves the naturalness in random sample generation. Furthermore, initial experiments demonstrate that randomly sampling from the proposed model can be used as data augmentation to improve the ASR performance.

* To appear in ICASSP 2020 

  Access Paper or Ask Questions

Improving Speech Translation by Understanding and Learning from the Auxiliary Text Translation Task

Jul 12, 2021
Yun Tang, Juan Pino, Xian Li, Changhan Wang, Dmitriy Genzel

Pretraining and multitask learning are widely used to improve the speech to text translation performance. In this study, we are interested in training a speech to text translation model along with an auxiliary text to text translation task. We conduct a detailed analysis to understand the impact of the auxiliary task on the primary task within the multitask learning framework. Our analysis confirms that multitask learning tends to generate similar decoder representations from different modalities and preserve more information from the pretrained text translation modules. We observe minimal negative transfer effect between the two tasks and sharing more parameters is helpful to transfer knowledge from the text task to the speech task. The analysis also reveals that the modality representation difference at the top decoder layers is still not negligible, and those layers are critical for the translation quality. Inspired by these findings, we propose three methods to improve translation quality. First, a parameter sharing and initialization strategy is proposed to enhance information sharing between the tasks. Second, a novel attention-based regularization is proposed for the encoders and pulls the representations from different modalities closer. Third, an online knowledge distillation is proposed to enhance the knowledge transfer from the text to the speech task. Our experiments show that the proposed approach improves translation performance by more than 2 BLEU over a strong baseline and achieves state-of-the-art results on the \textsc{MuST-C} English-German, English-French and English-Spanish language pairs.

* Accepted by ACL 2021 

  Access Paper or Ask Questions

Knowledge Transfer from Large-scale Pretrained Language Models to End-to-end Speech Recognizers

Feb 16, 2022
Yotaro Kubo, Shigeki Karita, Michiel Bacchiani

End-to-end speech recognition is a promising technology for enabling compact automatic speech recognition (ASR) systems since it can unify the acoustic and language model into a single neural network. However, as a drawback, training of end-to-end speech recognizers always requires transcribed utterances. Since end-to-end models are also known to be severely data hungry, this constraint is crucial especially because obtaining transcribed utterances is costly and can possibly be impractical or impossible. This paper proposes a method for alleviating this issue by transferring knowledge from a language model neural network that can be pretrained with text-only data. Specifically, this paper attempts to transfer semantic knowledge acquired in embedding vectors of large-scale language models. Since embedding vectors can be assumed as implicit representations of linguistic information such as part-of-speech, intent, and so on, those are also expected to be useful modeling cues for ASR decoders. This paper extends two types of ASR decoders, attention-based decoders and neural transducers, by modifying training loss functions to include embedding prediction terms. The proposed systems were shown to be effective for error rate reduction without incurring extra computational costs in the decoding phase.

* To be presented in ICASSP 2022 

  Access Paper or Ask Questions