Jump to Content
Ron J. Weiss

Ron J. Weiss

Authored Publications
Google Publications
Other Publications
Sort By
  • Title
  • Title, descending
  • Year
  • Year, descending
    Preview abstract We describe a sequence-to-sequence neural network which can directly generate speech waveforms from text inputs. The architecture extends the Tacotron model by incorporating a normalizing flow in the decoder loop. Output waveforms are modeled as a sequence of non-overlapping fixed-length frames, each one containing hundreds of samples. The inter-dependencies of waveform samples within each frame are modeled using the normalizing flow, enabling parallel training and synthesis. Longer-term dependencies are handled autoregressively by conditioning each flow on its preceding frames. The model allows for straightforward optimization towards the maximum likelihood objective, without utilizing intermediate spectral features nor additional loss terms. Contemporary state-of-the-art TTS systems use a sequence of separately learned models: one (such as Tacotron) which generates intermediate features (such as spectograms) from text, followed by a vocoder model (such as WaveRNN) which generates waveform samples from the intermediate features. The proposed system, in contrast, does not use a fixed intermediate representation ,and learns all parameters end-to-end. We demonstrate (to the best of our knowledge) the first system in the literature to do so successfully. Experiments show that the quality of speech generated from the proposed model is nearly competitive with the state-of-the-art neural TTS methods, with significantly improved generation speed. View details
    Preview abstract Although neural end-to-end text-to-speech models can synthesizehighly natural speech, there is still a room for improvements in itsefficiency during inference. This paper proposes a non-autoregressiveneural text-to-speech model augmented with a variational autoencoder-based residual encoder. This model, calledParallel Tacotron, is highlyparallelizable during both training and inference, allowing efficientsynthesis on modern parallel hardware. The use of the variationalautoencoder helps to relax the one-to-many mapping nature of thetext-to-speech problem. To further improve the naturalness, weintroduce an iterative spectrogram loss, which is inspired by iterativerefinement, and lightweight convolution, which can efficiently capturelocal contexts. Experimental results show that Parallel Tacotronmatches a strong autoregressive baseline in subjective naturalnesswith significantly decreased inference time. View details
    Preview abstract This paper introduces WaveGrad, a conditional model for waveform generation which estimates gradients of the data density. The model is built on prior work on score matching and diffusion probabilistic models. It starts from a Gaussian white noise signal and iteratively refines the signal via a gradient-based sampler conditioned on the mel-spectrogram. WaveGrad offers a natural way to trade inference speed for sample quality by adjusting the number of refinement steps, and bridges the gap between non-autoregressive and autoregressive models in terms of audio quality. We find that it can generate high fidelity audio samples using as few as six iterations. Experiments reveal WaveGrad to generate high fidelity audio, outperforming adversarial non-autoregressive baselines and matching a strong likelihood-based autoregressive baseline using fewer sequential operations. Audio samples are available at https://wavegrad.github.io/ View details
    WaveGrad 2: Iterative Refinement for Text-to-Speech Synthesis
    Nanxin Chen
    Yu Zhang
    Mohammad Norouzi
    Najim Dehak
    William Chan
    Interspeech (2021)
    Preview abstract This paper introduces WaveGrad 2, an end-to-end non-autoregressive generative model for text-to-speech synthesis trained to estimate the gradients of the data density. Unlike recent TTS systems which are a cascade of separately learned models, during training the proposed model requires only text or phoneme sequence, learns all parameters end-to-end without intermediate features, and can generate natural speech audio with great varieties. This is achieved by the score matching objective, which optimizes the network to model the score function of the real data distribution. Output waveforms are generated using an iterative refinement process beginning from a random noise sample. Like our prior work, WaveGrad 2 offers a natural way to trade inference speed for sample quality by adjusting the number of refinement steps. Experiments reveal that the model can generate high fidelity audio, closing the gap between end-to-end and contemporary systems, approaching the performance of a state-of-the-art neural TTS system. We further carry out various ablations to study the impact of different model configurations. View details
    Preview abstract We propose a multitask training method for attention-basedend-to-end speech recognition models. We regularize the de-coder in a listen, attend, and spell model by multitask trainingon both audio-text and text-only data. Trained on the 100-hoursubset of LibriSpeech, the proposed method leads to an 11%relative performance improvement over the baseline and is com-parable to language model shallow fusion, without requiring anadditional neural network during decoding. We observe a simi-lar trend on the whole 960-hour LibriSpeech training set. Anal-yses of sample output sentences demonstrate that the proposedmethod can incorporate language level information, suggestingits effectiveness in real-world applications View details
    Preview abstract Supervised neural network training has led to significant progress on single-channel sound separation. This approach relies on ground truth isolated sources, which precludes scaling to widely available mixture data and limits progress on open-domain tasks. The recent mixture invariant training (MixIT) method enables training on in-the-wild data; however, it suffers from two outstanding problems. First, it produces models which tend to over-separate, producing more output sources than are present in the input. Second, the exponential computational complexity of the MixIT loss limits the number of feasible output sources. In this paper we address both issues. To combat over-separation we introduce new losses: sparsity losses that favor fewer output sources and a covariance loss that discourages correlated outputs. We also experiment with a semantic classification loss by predicting weak class labels for each mixture. To handle larger numbers of sources, we introduce an efficient approximation using a fast least-squares solution, projected onto the MixIT constraint set. Our experiments show that the proposed losses curtail over-separation and improve overall performance. The best performance is achieved using larger numbers of output sources, enabled by our efficient MixIT loss, combined with sparsity losses to prevent over-separation. On the FUSS test set, we achieve over 13 dB in multi-source SI-SNR improvement, while boosting single-source reconstruction SI-SNR by over 17 dB. View details
    Preview abstract We propose a hierarchical, fine-grained and interpretable latent model for prosody based on the Tacotron~2. This model achieves multi-resolution modeling by conditioning finer level prosody representations on coarser level ones. In addition, the hierarchical conditioning is also imposed across all latent dimensions using a conditional VAE structure which exploits an auto-regressive structure. Reconstruction performance is evaluated with the $F_0$ frame error (FFE) and the mel-cepstral distortion (MCD) which illustrates the new structure does not degrade the model. Interpretations of prosody attributes are provided together with the comparison between word-level and phone-level prosody representations. Moreover, both qualitative and quantitative evaluations are used to demonstrate the improvement in the disentanglement of the latent dimensions. View details
    Preview abstract In recent years, rapid progress has been made on the problem of single-channel sound separation using supervised training of deep neural networks. In such supervised approaches, a model is trained to predict the component sources from synthetic mixtures created by adding up isolated ground-truth sources. Reliance on this synthetic training data is problematic because good performance depends upon the degree of match between the training data and real-world audio, especially in terms of the acoustic conditions and distribution of sources. The acoustic properties can be challenging to accurately simulate, and the distribution of sound types may be hard to replicate. In this paper, we propose a completely unsupervised method, mixture invariant training (MixIT), that requires only single-channel acoustic mixtures. In MixIT, training examples are constructed by mixing together existing mixtures, and the model separates them into a variable number of latent sources, such that the separated sources can be remixed to approximate the original mixtures. We show that MixIT can achieve competitive performance compared to supervised methods on speech separation. Using MixIT in a semi-supervised learning setting enables unsupervised domain adaptation and learning from large amounts of real world data without ground-truth source waveforms. In particular, we significantly improve reverberant speech separation performance by incorporating reverberant mixtures, train a speech enhancement system from noisy mixtures, and improve universal sound separation by incorporating a large amount of in-the-wild data. View details
    Preview abstract Recently, we introduced a 2-pass on-device E2E model, which runs RNN-T in the first-pass and then rescores/redecodes this with a LAS decoder. This on-device model was similar in performance compared to a state-of-the-art conventional model. However, like many E2E models it is trained on supervised audio-text pairs and thus did poorly on rare-words compared to a conventional model trained on a much larger text-corpora. In this work, we introduce a joint acoustic and text-only decoder (JATD) into the LAS decoder, which allows the LAS decoder to be trained on a much larger text-corporate. We find that the JATD model provides between a 3-10\% relative improvement in WER compared to a LAS decoder trained on only supervised audio-text pairs across a variety of proper noun test sets. View details
    Preview abstract Recently proposed approaches for fine-grained prosody control of end-to-end text-to-speech samples enable precise control of the prosody of synthesized speech. Such models incorporate a fine-grained variational autoencoder (VAE) structure into a sequence-to-sequence model, extracting latent prosody features for each input token (e.g.\ phonemes). Generating samples using the standard VAE prior, an independent Gaussian at each time step, results in very unnatural and discontinuous speech, with dramatic variation between phonemes. In this paper we propose a sequential prior in the discrete latent space which can be used to generate more natural samples. This is accomplished by discretizing the latent prosody features using vector quantization, and training an autoregressive (AR) prior model over the result. The AR prior is learned separately from the training of the posterior. We evaluate the approach using subjective listening tests, objective metrics of automatic speech recognition (ASR) performance, as well as measurements of prosody attributes including volume, pitch, and phoneme duration. Compared to the fine-grained VAE baseline, the proposed model achieves equally good copy synthesis reconstruction performance, but significantly improves naturalness in sample generation. The diversity of the prosody in random samples better matches that of the real speech. Furthermore, initial experiments demonstrate that samples generated from the quantized latent sapce can be used as an effective data augmentation strategy to improve ASR performance. View details
    Preview abstract Supervised approaches to single-channel speech separation rely on synthetic mixtures, so that the individual sources can be used as targets. Good performance depends upon how well the synthetic mixture data match real mixtures. However, matching synthetic data to the acoustic properties and distribution of sounds in a target domain can be challenging. Instead, we propose an unsupervised method that requires only singlechannel acoustic mixtures, without ground-truth source signals. In this method, existing mixtures are mixed together to form a mixture of mixtures, which the model separates into latent sources. We propose a novel loss that allows the latent sources to be remixed to approximate the original mixtures. Experiments show that this method can achieve competitive performance on speech separation compared to supervised methods. In a semisupervised learning setting, our method enables domain adaptation by incorporating unsupervised mixtures from a matched domain. In particular, we demonstrate that significant improvement to reverberant speech separation performance can be achieved by incorporating reverberant mixtures. View details
    Preview abstract Texture synthesis techniques based on matching the Gram matrix of feature activations in neural networks have achieved spectacular success in the image domain. In this paper we extend these techniques to the audio domain. We demonstrate that synthesizing diverse audio textures is challenging, and argue that this is because audio data is relatively low-dimensional. We therefore introduce two new terms to the original Grammian loss: an autocorrelation term that preserves rhythm, and a diversity term that encourages the optimization procedure to synthesize unique textures. We quantitatively study the impact of our design choices on the quality of the synthesized audio by introducing an audio analogue to the Inception loss which we term the VGGish loss. We show that there is a trade-off between the diversity and quality of the synthesized audio using this technique. Finally we perform a number of experiments to qualitatively study how these design choices impact the quality of the synthesized audio. View details
    Preview abstract We present a multispeaker, multilingual text-to-speech (TTS) synthesis model based on Tacotron that is able to produce high quality speech in multiple languages. Moreover, the model is able to transfer voices across languages, e.g. synthesize fluent Spanish speech using an English speaker's voice, without training on any bilingual or parallel examples. Such transfer works across distantly related languages, e.g. English and Mandarin. Critical to achieving this result are: 1. using a phonemic input representation to encourage sharing of model capacity across languages, and 2. incorporating an adversarial loss term to encourage the model to disentangle its representation of speaker identity (which is perfectly correlated with language in the training data) from the speech content. Further scaling up the model by training on multiple speakers of each language, and incorporating an autoencoding input to help stabilize attention during training, results in a model which can be used to consistently synthesize intelligible speech for training speakers in all languages seen during training, and in native or foreign accents. View details
    Hierarchical Generative Modeling for Controllable Speech Synthesis
    Wei-Ning Hsu
    Yu Zhang
    Yuxuan Wang
    Ye Jia
    Jonathan Shen
    Patrick Nguyen
    Ruoming Pang
    International Conference on Learning Representations (2019)
    Preview abstract This paper proposes a neural end-to-end text-to-speech model which can control latent attributes in the generation of speech, that are rarely annotated in the training data (e.g. speaking styles, accents, background noise level, and recording conditions). The model is formulated as a conditional generative model with two levels of hierarchical latent variables. The first level is a categorical variable, which represents attribute groups (e.g. clean/noisy) and provides interpretability. The second level, conditioned on the first, is a multivariate Gaussian variable, which characterizes specific attribute configurations (e.g. noise level, speaking rate) and enables disentangled fine-grained control over these attributes. This amounts to using a Gaussian mixture model (GMM) for the latent distribution. Extensive evaluation of the proposed model demonstrates its ability to control the aforementioned attributes. In particular, it is capable of consistently synthesizing high-quality clean speech regardless of the quality of the training data for the target speaker. View details
    Preview abstract This paper introduces a new speech corpus called ``LibriTTS'' designed for text-to-speech use. It is derived from the original audio and text materials of the LibriSpeech corpus, which has been used for training and evaluating automatic speech recognition systems. The new corpus inherits desired properties of the LibriSpeech corpus while addressing a number of issues which make LibriSpeech less than ideal for text-to-speech work. The released corpus consists of 585 hours of speech data at 24kHz sampling rate from 2,456 speakers and the corresponding texts. Experimental results show that neural end-to-end TTS models trained from the LibriTTS corpus achieved above 4.0 in mean opinion scores in naturalness in five out of six evaluation speakers. The corpus is freely available for download from http://www.openslr.org/60/. View details
    Preview abstract End-to-end Speech Translation (ST) models have many potential advantages when compared to the cascade of Automatic Speech Recognition (ASR) and text Machine Translation (MT) models, including lowered inference latency and the avoidance of error compounding. However, the quality of end-to-end ST is often limited by a paucity of training data, since it is difficult to collect large parallel corpora of speech and translated transcript pairs. Previous studies have proposed the use of pre-trained components and multi-task learning in order to benefit from weakly supervised training data, such as speech-totranscript or text-to-foreign-text pairs. In this paper, we demonstrate that using pre-trained MT or text-to-speech (TTS) synthesis models to convert weakly supervised data into speech-to-translation pairs for ST training can be more effective than multi-task learning. Furthermore, we demonstrate that a high quality end-to-end ST model can be trained using only weakly supervised datasets, and that synthetic data sourced from unlabeled monolingual text or speech can be used to improve performance. Finally, we discuss methods for avoiding overfitting to synthetic speech with a quantitative ablation study. View details
    Preview abstract We describe Parrotron, an end-to-end-trained speech-to-speech conversion model that maps an input spectrogram directly to another spectrogram, without utilizing any intermediate discrete representation. The network is composed of an encoder, spectrogram and phoneme decoders, followed by a vocoder to synthesize a time-domain waveform. We demonstrate that this model can be trained to normalize speech from any speaker regardless of accent, prosody, and background noise, into the voice of a single canonical target speaker with a fixed accent and consistent articulation and prosody. We further show that this normalization model can be adapted to normalize highly atypical speech from a deaf speaker, resulting in significant improvements in intelligibility and naturalness, measured via a speech recognizer and listening tests. Finally, demonstrating the utility of this model on other speech tasks, we show that the same model architecture can be trained to perform a speech separation task View details
    Preview abstract We present an attention-based sequence-to-sequence neural network which can directly translate speech from one language into speech in another language, without relying on an intermediate text representation. The network is trained end-to-end, learning to map speech spectrograms into target spectrograms in another language, corresponding to the translated content (in a different canonical voice). We further demonstrate the ability to synthesize translated speech using the voice of the source speaker. We conduct experiments on two Spanish-to-English speech translation datasets, and find that the proposed model slightly underperforms a baseline cascade of a direct speech-to-text translation model and a text-to-speech synthesis model, demonstrating the feasibility of the approach on this very challenging task. View details
    Preview abstract Attention-based sequence-to-sequence models for speech recognition jointly train an acoustic model, language model (LM), and alignment mechanism using a single neural network and require only parallel audio-text pairs. Thus, the language model component of the end-to-end model is only trained on transcribed audio-text pairs, which leads to performance degradation especially on rare words. While there have been a variety of work that look at incorporating an external LM trained on text-only data into the end-to-end framework, none of them have taken into account the characteristic error distribution made by the model. In this paper, we propose a novel approach to utilizing text-only data, by training a spelling correction (SC) model to explicitly correct those errors. On the LibriSpeech dataset, we demonstrate that the proposed model results in an 18.6\% relative improvement in WER over the baseline model when directly correcting top ASR hypothesis, and a 29.0\% relative improvement when further rescoring an expanded n-best list using an external LM. View details
    Preview abstract To leverage crowd-sourced data to train multi-speaker text-to-speech (TTS) models that can synthesize clean speech for all speakers, it is essential to learn disentangled representations which can independently control the speaker identity and background noise in generated signals. However, learning such representations can be challenging, due to the lack of labels describing the recording conditions of each training example, and the fact that speakers and recording conditions are often correlated, e.g. since users often make many recordings using the same equipment. This paper proposes three components to address this problem by: (1) formulating a conditional generative model with factorized latent variables, (2) using data augmentation to add noise that is not correlated with speaker identity and whose label is known during training, and (3) using adversarial factorization to improve disentanglement. Experimental results demonstrate that the proposed method can disentangle speaker and noise attributes even if they are correlated in the training data, and can be used to consistently synthesize clean speech for all speakers. Ablation studies verify the importance of each proposed component. View details
    Unsupervised speech representation learning using WaveNet autoencoders
    Jan Chorowski
    Samy Bengio
    Aäron van den Oord
    IEEE Transactions on Audio, Speech, and Language Processing (2019)
    Preview abstract We consider the task of unsupervised extraction of meaningful latent representations of speech by applying auto-encoding neural networks to speech waveforms. The goal is to learn a representation which is able to capture high level semantic content from the signal, e.g. phoneme identities, while being invariant to confounding low level details in the signal such as the underlying pitch contour or background noise. The behavior of auto-encoder models depends on the kind of constraint that is applied to the latent representation. We compare three variants: a simple dimensionality reduction bottleneck, a Gaussian Variational Auto-Encoder (VAE), and a discrete Vector Quantized VAE (VQ-VAE). We analyze the quality of the learned representation in terms of its speaker independence, the ability to predict phonetic content, and the ability to accurately reconstruct individual spectrogram frames. Moreover, for the discrete encodings extracted using the VQ-VAE, we measure the ease of mapping them to phonemes. We introduce a regularization scheme that forces the representations to concentrate on the phonetic content of the utterance and report performance comparable with the top entries in the ZeroSpeech 2017 unsupervised acoustic unit discovery task. View details
    Preview abstract Inspired by recent work on neural network image generation which rely on backpropagation towards the network inputs, we present a proof-of-concept system for speech texture synthesis and voice conversion based on two mechanisms: approximate inversion of the representation learned by a speech recognition neural network, and on matching statistics of neuron activations between different source and target utterances. Similar to image texture synthesis and neural style transfer, the system works by optimizing a cost function with respect to the input waveform samples. To this end we use a differentiable mel-filterbank feature extraction pipeline and train a convolutional CTC speech recognition network. Our system is able to extract speaker characteristics from very limited amounts of target speaker data, as little as a few seconds, and can be used to generate realistic speech babble or reconstruct an utterance in a different voice. View details
    Preview abstract Training a conventional automatic speech recognition (ASR) system to support multiple languages is challenging because the sub-word unit, lexicon and word inventories are typically language specific. In contrast, sequence-to-sequence models are well suited for multilingual ASR because they encapsulate an acoustic, pronunciation and language model jointly in a single network. In this work we present a single sequence-to-sequence ASR model trained on 9 different Indian languages, which have very little overlap in their scripts. Specifically, we take a union of language-specific grapheme sets and train a grapheme-based sequence-to-sequence model jointly on data from all languages. We find that this model, which is not explicitly given any information about language identity, improves recognition performance by 21% relative compared to analogous sequence-to-sequence models trained on each language individually. By modifying the model to accept a language identifier as an additional input feature, we further improve performance by an additional 7% relative and eliminate confusion between different languages. View details
    Preview abstract We describe a neural network-based system for text-to-speech (TTS) synthesis that is able to generate speech audio in the voice of many different speakers, including those unseen during training. Our system consists of three independently trained components: (1) a speaker encoder network, trained on a speaker verification task using an independent dataset of noisy speech from thousands of speakers without transcripts, to generate a fixed-dimensional embedding vector from seconds of reference speech from a target speaker; (2) a sequence-to-sequence synthesis network based on Tacotron 2, which generates a mel spectrogram from text, conditioned on the speaker embedding; (3) an auto-regressive WaveNet-based vocoder that converts the mel spectrogram into a sequence of time domain waveform samples. We demonstrate that the proposed model is able to transfer the knowledge of speaker variability learned by the discriminatively-trained speaker encoder to the new task, and is able to synthesize natural speech from speakers that were not seen during training. We quantify the importance of training the speaker encoder on a large and diverse speaker set in order to obtain the best generalization performance. Finally, we show that randomly sampled speaker embeddings can be used to synthesize speech in the voice of novel speakers dissimilar from those used in training, indicating that the model has learned a high quality speaker representation. View details
    Preview abstract We present an extension to the Tacotron speech synthesis architecture that learns a latent embedding space of prosody, derived from a reference acoustic representation containing the desired prosody. We show that conditioning Tacotron on this learned embedding space results in synthesized audio that matches the reference signal’s prosody with fine time detail. We define several quantitative and subjective metrics for evaluating prosody transfer, and report results and audio samples from a single-speaker and 44-speaker Tacotron model on a prosody transfer task. View details
    Natural TTS Synthesis By Conditioning WaveNet On Mel Spectrogram Predictions
    Jonathan Shen
    Ruoming Pang
    Mike Schuster
    Navdeep Jaitly
    Zongheng Yang
    Yu Zhang
    Yuxuan Wang
    Yannis Agiomyrgiannakis
    ICASSP (2018)
    Preview abstract This paper describes Tacotron 2, a neural network architecture for speech synthesis directly from text. The system is composed of a recurrent sequence-to-sequence feature prediction network that maps character embeddings to mel-scale spectrograms, followed by a modified WaveNet model acting as a vocoder to synthesize timedomain waveforms from those spectrograms. Our model achieves a mean opinion score (MOS) of 4.53 comparable to a MOS of 4.58 for professionally recorded speech. To validate our design choices, we present ablation studies of key components of our system and evaluate the impact of using mel spectrograms as the input to WaveNet instead of linguistic, duration, and F0 features. We further demonstrate that using a compact acoustic intermediate representation enables significant simplification of the WaveNet architecture. View details
    Preview abstract In this paper, we present a novel system that separates the voice of a target speaker from multi-speaker signals, by making use of a reference signal from the target speaker. We achieve this by training two separate neural networks: (1) A speaker recognition network that produces speaker-discriminative embeddings; (2) A spectrogram masking network that takes both noisy spectrogram and speaker embedding as input, and produces a mask. Our system significantly reduces the speech recognition WER on multi-speaker signals, with minimal WER degradation on single-speaker signals. View details
    Preview abstract Attention-based encoder-decoder architectures such as Listen, Attend, and Spell (LAS), subsume the acoustic, pronunciation and language model components of a traditional automatic speech recognition (ASR) system into a single neural network. In our previous work, we have shown that such architectures are comparable to state-of-the-art ASR systems on dictation tasks, but it was not clear if such architectures would be practical for more challenging tasks such as voice search. In this work, we explore a variety of structural and optimization improvements to our LAS model which significantly improve performance. On the structural side, we show that word piece models can be used instead of graphemes. We introduce a multi-head attention architecture, which offers improvements over the commonly-used single-head attention. On the optimization side, we explore techniques such as synchronous training, scheduled sampling, label smoothing, and minimum word error rate optimization, which are all shown to improve accuracy. We present results with a unidirectional LSTM encoder for streaming recognition. On a 12,500 hour voice search task, we find that the proposed changes improve the WER of the LAS system from 9.2% to 5.6%, while the best conventional system achieve 6.7% WER. We also test both models on a dictation dataset, and our model provide 4.1% WER while the conventional system provides 5% WER. View details
    Preview abstract We present a recurrent encoder-decoder deep neural network architecture that directly translates speech in one language into text in another. The model does not explicitly transcribe the speech into text in the source language, nor does it require supervision from the ground truth source language transcription during training. We apply a slightly modified sequence-to-sequence with attention architecture that has previously been used for speech recognition and show that it can be repurposed for this more complex task, illustrating the power of attention-based models. A single model trained end-to-end obtains state-of-the-art performance on the Fisher Callhome Spanish-English speech translation task, outperforming a cascade of independently trained sequence-to-sequence speech recognition and machine translation models by 1.8 BLEU points on the Fisher test set. In addition, we find that making use of the training data in both languages by multi-task training sequence-to-sequence speech translation and recognition models with a shared encoder network can improve performance by a further 1.4 BLEU points. View details
    Preview abstract Convolutional Neural Networks (CNNs) have proven very effective in image classification and have shown promise for audio classification. We apply various CNN architectures to audio and investigate their ability to classify videos with a very large scale data set of 70M training videos (5.24 million hours) with 30,871 labels. We examine fully connected Deep Neural Networks (DNNs), AlexNet [1], VGG [2], Inception [3], and ResNet [4]. We explore the effects of training with different sized subsets of the 70M training videos. Additionally we report the effect of training over different subsets of the 30,871 labels. While our dataset contains video-level labels, we are also interested in Acoustic Event Detection (AED) and train a classifier on embeddings learned from the video-level task on AudioSet [5]. We find that derivatives of image classification networks do well on our audio classification task, that increasing the number of labels we train on provides some improved performance over subsets of labels, that performance of models improves as we increase training set size, and that a model using embeddings learned from the video-level task do much better than a baseline on the AudioSet classification task. View details
    Preview abstract This paper describes the technical and system building advances made to the Google Home multichannel speech recognition system, which was launched in November 2016. Technical advances include an adaptive dereverberation frontend, the use of neural network models that do multichannel processing jointly with acoustic modeling, and grid lstms to model frequency variations. On the system level, improvements include adapting the model using Google Home specific data. We present results on a variety of multichannel sets. The combination of technical and system advances result in a reduction of WER of over 18\% relative compared to the current production system. View details
    Preview abstract Multichannel ASR systems commonly separate speech enhancement, including localization, beamforming and postfiltering, from acoustic modeling. In this paper, we perform multichannel enhancement jointly with acoustic modeling in a deep neural network framework. Inspired by beamforming, which leverages differences in the fine time structure of the signal at different microphones to filter energy arriving from different directions, we explore modeling the raw time-domain waveform directly. We introduce a neural network architecture which performs multichannel filtering in the first layer of the network and show that this network learns to be robust to varying target speaker direction of arrival, performing as well as a model that is given oracle knowledge of the true target speaker direction. % Next, we show how performance can be improved by \emph{factoring} the first layer to separate the multichannel spatial filtering operation from a single channel filterbank which computes a frequency decomposition. % We also introduce an adaptive variant, which updates the spatial filter coefficients at each time frame based on the previous inputs. % Finally we demonstrate that these approaches can be implemented more efficiently in the frequency domain. Overall, we find that such multichannel neural networks give a relative word error rate improvement of more than 5\% compared to a traditional beamforming-based multichannel ASR system and more than 10\% compared to a single channel waveform model. View details
    Preview abstract Recurrent neural network models with an attention mechanism have proven to be extremely effective on a wide variety of sequence-to-sequence problems. However, the fact that soft attention mechanisms perform a pass over the entire input sequence when producing each element in the output sequence precludes their use in online settings and results in a quadratic time complexity. Based on the insight that the alignment between input and output sequence elements is monotonic in many problems of interest, we propose an end-to-end differentiable method for learning monotonic alignments which, at test time, enables computing attention online and in linear time. We validate our approach on sentence summarization, machine translation, and online speech recognition problems and achieve results competitive with existing sequence-to-sequence models. View details
    Preview abstract A text-to-speech synthesis system typically consists of multiple stages, such as a text analysis frontend, an acoustic model and an audio synthesis module. Building these components often requires extensive domain expertise and may contain brittle design choices. In this paper, we present Tacotron, an end-to-end generative text-to-speech model that synthesizes speech directly from characters. Given (text, audio) pairs, the model can be trained completely from scratch with random initialization. We present several key techniques to make the sequence-to-sequence framework perform well for this challenging task. Tacotron achieves a 3.82 subjective 5-scale mean opinion score on US English, outperforming a production parametric system in terms of naturalness. In addition, since Tacotron generates speech at the frame level, it's substantially faster than sample-level autoregressive methods. View details
    Preview abstract Multichannel ASR systems commonly separate speech enhancement, including localization, beamforming and postfiltering, from acoustic modeling. In this chapter, we perform multi-channel enhancement jointly with acoustic modeling in a deep neural network framework. Inspired by beamforming, which leverages differences in the fine time structure of the signal at different microphones to filter energy arriving from different directions, we explore modeling the raw time-domain waveform directly. We introduce a neural network architecture which performs multichannel filtering in the first layer of the network and show that this network learns to be robust to varying target speaker direction of arrival, performing as well as a model that is given oracle knowledge of the true target speaker direction. Next, we show how performance can be improved by factoring the first layer to separate the multichannel spatial filtering operation from a single channel filterbank which computes a frequency decomposition. We also introduce an adaptive variant, which updates the spatial filter coefficients at each time frame based on the previous inputs. Finally we demonstrate that these approaches can be implemented more efficiently in the frequency domain. Overall, we find that such multichannel neural networks give a relative word error rate improvement of more than 5% compared to a traditional beamforming-based multichannel ASR system and more than 10% compared to a single channel waveform model. View details
    Preview abstract Joint multichannel enhancement and acoustic modeling using neural networks has shown promise over the past few years. However, one shortcoming of previous work [1,2,3] is that the filters learned during training are fixed for decoding, potentially limiting the ability of these models to adapt to previously unseen or changing conditions. In this paper we explore a neural network adaptive beamforming (NAB) technique to address this issue. Specifically, we use LSTM layers to predict time domain beamforming filter coefficients at each input frame. These filters are convolved with the framed time domain input signal and summed across channels, essentially performing FIR filter-and-sum beamforming using the dynamically adapted filter. The beamformer output is passed into a waveform CLDNN acoustic model [4] which is trained jointly with the filter prediction LSTM layers. We find that the proposed NAB model achieves a 12.7% relative improvement in WER over a single channel model [4] and reaches similar performance to a ``factored'' model architecture which utilizes several fixed spatial filters [3] on a 2,000-hour Voice Search task, with a 17.9% decrease in computational cost. View details
    Speech Acoustic Modeling from Raw Multichannel Waveforms
    Yedid Hoshen
    International Conference on Acoustics, Speech, and Signal Processing, IEEE (2015)
    Preview abstract Standard deep neural network-based acoustic models for automatic speech recognition (ASR) rely on hand-engineered input features, typically log-mel filterbank magnitudes. In this paper, we describe a convolutional neural network - deep neural network (CNN-DNN) acoustic model which takes raw multichannel waveforms as input, i.e. without any preceding feature extraction, and learns a similar feature representation through supervised training. By operating directly in the time domain, the network is able to take advantage of the signal's fine time structure that is discarded when computing filterbank magnitude features. This structure is especially useful when analyzing multichannel inputs, where timing differences between input channels can be used to localize a signal in space. The first convolutional layer of the proposed model naturally learns a filterbank that is selective in both frequency and direction of arrival, i.e. a bank of bandpass beamformers with an auditory-like frequency scale. When trained on data corrupted with noise coming from different spatial locations, the network learns to filter them out by steering nulls in the directions corresponding to the noise sources. Experiments on a simulated multichannel dataset show that the proposed acoustic model outperforms a DNN that uses log-mel filterbank magnitude features under noisy and reverberant conditions. View details
    Affinity Weighted Embedding
    Jason Weston
    International Conference on Machine Learning (2014)
    Preview abstract Supervised linear embedding models like Wsabie (Weston et al., 2011) and supervised semantic indexing (Bai et al., 2010) have proven successful at ranking, recommendation and annotation tasks. However, despite being scalable to large datasets they do not take full advantage of the extra data due to their linear nature, and we believe they typically underfit. We propose a new class of models which aim to provide improved performance while retaining many of the benefits of the existing class of embedding models. Our approach works by reweighting each component of the embedding of features and labels with a potentially nonlinear affinity function. We describe several variants of the family, and show its usefulness on several datasets. View details
    Affinity Weighted Embedding
    Jason Weston
    International Conference on Learning Representations (2013)
    Nonlinear Latent Factorization by Embedding Multiple User Interests
    Jason Weston
    ACM International Conference on Recommender Systems (RecSys) (2013)
    Preview abstract Classical matrix factorization approaches to collaborative filtering learn a latent vector for each user and each item, and recommendations are scored via the similarity between two such vectors, which are of the same dimension. In this work, we are motivated by the intuition that a user is a much more complicated entity than any single item, and cannot be well described by the same representation. Hence, the variety of a user’s interests could be better captured by a more complex representation. We propose to model the user with a richer set of functions, specifically via a set of latent vectors, where each vector captures one of the user’s latent interests or tastes. The overall recommendation model is then nonlinear where the matching score between a user and a given item is the maximum matching score over each of the user’s latent interests with respect to the item’s latent representation. We describe a simple, general and efficient algorithm for learning such a model, and apply it to large scale, real world datasets from YouTube and Google Music, where our approach outperforms existing techniques. View details
    Learning to Rank Recommendations with the k-Order Statistic Loss
    Jason Weston
    ACM International Conference on Recommender Systems (RecSys) (2013)
    Preview abstract Making recommendations by learning to rank is becoming an increasingly studied area. Approaches that use stochastic gradient descent scale well to large collaborative filtering datasets, and it has been shown how to approximately optimize the mean rank, or more recently the top of the ranked list. In this work we present a family of loss functions, the korder statistic loss, that includes these previous approaches as special cases, and also derives new ones that we show to be useful. In particular, we present (i) a new variant that more accurately optimizes precision at k, and (ii) a novel procedure of optimizing the mean maximum rank, which we hypothesize is useful to more accurately cover all of the user’s tastes. The general approach works by sampling N positive items, ordering them by the score assigned by the model, and then weighting the example as a function of this ordered set. Our approach is studied in two real-world systems, Google Music and YouTube video recommendations, where we obtain improvements for computable metrics, and in the YouTube case, increased user click through and watch duration when deployed live on www.youtube.com. View details
    Latent Collaborative Retrieval
    Jason Weston
    Chong Wang
    Adam Berenzweig
    International Conference on Machine Learning (2012)
    Preview abstract Retrieval tasks typically require a ranking of items given a query. Collaborative filtering tasks, on the other hand, learn models comparing users with items. In this paper we study the joint problem of recommending items to a user with respect to a given query, which is a surprisingly common task. This setup differs from the standard collaborative filtering one in that we are given a query × user × item tensor for training instead of the more traditional user × item matrix. Compared to document retrieval we do have a query, but we may or may not have content features (we will consider both cases) and we can also take account of the user’s profile. We introduce a factorized model for this new task that optimizes the top ranked items returned for the given query and user. We report empirical results where it outperforms several baselines. View details
    No Results Found