Apple sponsored the 45th International Conference on Acoustics, Speech, and Signal Processing (ICASSP) in May 2020. With a focus on signal processing and its applications, the conference took place virtually from May 4 - 8. Read Apple’s accepted papers below.

Learn more about ICASSP

See our teams at Jobs at Apple

Accepted Papers

Detecting Emotion Primitives From Speech And Their Use In Discerning Categorical Emotions

Vasudha Kowtha, Vikramjit Mitra, Chris Bartels, Erik Marchi, Sue Booker, William Caruso, Sachin Kajarekar, Devang Naik

Emotion plays an essential role in human-to-human communication, enabling us to convey feelings such as happiness, frustration, and sincerity. While modern speech technologies rely heavily on speech recognition and natural language understanding for speech content understanding, the investigation of vocal expression is increasingly gaining attention. Key considerations for building robust emotion models include characterizing and improving the extent to which a model, given its training data distribution, is able to generalize to unseen data conditions. This work investigated a long-shot-term memory (LSTM) network and a time convolution - LSTM (TC-LSTM) to detect primitive emotion attributes such as valence, arousal, and dominance, from speech. It was observed that training with multiple datasets and using robust features improved the concordance correlation coefficient (CCC) for valence, by 30% with respect to the baseline system. Additionally, this work investigated how emotion primitives can be used to detect categorical emotions such as happiness, disgust, contempt, anger, and surprise from neutral speech, and results indicated that arousal, followed by dominance was a better detector of such emotions.

Improving Language Identification for Multilingual Speakers

Andrew Titus, Jan Silovsky, Nanxin Chen, Roger Hsiao, Mary Young, Arnab Ghoshal

Spoken language identification (LID) technologies have improved in recent years from discriminating largely distinct languages to discriminating highly similar languages or even dialects of the same language. One aspect that has been mostly neglected, however, is discrimination of languages for multilingual speakers, despite being a primary target audience of many systems that utilize LID technologies. As we show in this work, LID systems can have a high average accuracy for most combinations of languages while greatly underperforming for others when accented speech is present. We address this by using coarser-grained targets for the acoustic LID model and integrating its outputs with interaction context signals in a context-aware model to tailor the system to each user. This combined system achieves an average 97% accuracy across all language combinations while improving worst-case accuracy by over 60% relative to our baseline.

Lattice-Based False Trigger Mitigation Using Graph Neural Networks

Pranay Dighe, Saurabh Adya, Nuoyu Li, Srikanth Vishnubhotla, Devang Naik, Adithya Sagar, Ying Ma, Stephen Pulman, Jason Williams

Voice-triggered smart assistants often rely on detection of a trigger-phrase before they start listening for the user request. Mitigation of false triggers is an important aspect of building a privacy-centric non-intrusive smart assistant. In this paper, we address the task of false trigger mitigation (FTM) using a novel approach based on analyzing automatic speech recognition (ASR) lattices using graph neural networks (GNN). The proposed approach uses the fact that decoding lattice of a falsely triggered audio exhibits uncertainties in terms of many alternative paths and unexpected words on the lattice arcs as compared to the lattice of a correctly triggered audio. A pure trigger-phrase detector model doesn't fully utilize the intent of the user speech whereas by using the complete decoding lattice of user audio, we can effectively mitigate speech not intended for the smart assistant. We deploy two variants of GNNs in this paper based on 1) graph convolution layers and 2) self-attention mechanism respectively. Our experiments demonstrate that GNNs are highly accurate in FTM task by mitigating ~87% of false triggers at 99% true positive rate (TPR). Furthermore, the proposed models are fast to train and efficient in parameter requirements.

Multi-task Learning for Voice Trigger Detection

Siddharth Sigtia, Pascal Clark, Rob Haynes, Hywel Richards, John Bridle

We describe the design of a voice trigger detection system for smart speakers. In this study, we address two major challenges. The first is that the detectors are deployed in complex acoustic environments with external noise and loud playback by the device itself. Secondly, collecting training examples for a specific keyword or trigger phrase is challenging resulting in a scarcity of trigger phrase specific training data. We describe a two-stage cascaded architecture where a low-power detector is always running and listening for the trigger phrase. If a detection is made at this stage, the candidate audio segment is re-scored by larger, more complex models to verify that the segment contains the trigger phrase. In this study, we focus our attention on the architecture and design of these second-pass detectors. We start by training a general acoustic model that produces phonetic transcriptions given a large labelled training dataset. Next, we collect a much smaller dataset of examples that are challenging for the baseline system. We then use multi-task learning to train a model to simultaneously produce accurate phonetic transcriptions on the larger dataset and discriminate between true and easily confusable examples using the smaller dataset. Our results demonstrate that the proposed model reduces errors by half compared to the baseline in a range of challenging test conditions without requiring extra parameters.

Generating Multilingual Voices using Speaker Space Translation Based on Bilingual Speaker Data

Soumi Maiti, Erik Marchi, Alistair Conkie

We present progress towards bilingual Text-to-Speech which is able to transform a monolingual voice to speak a second language while preserving speaker voice quality. We demonstrate that a bilingual speaker embedding space contains a separate distribution for each language and that a simple transform in speaker space generated by the speaker embedding can be used to control the degree of accent of a synthetic voice in a language. The same transform can be applied even to monolingual speakers. In our experiments speaker data from an English-Spanish (Mexican) bilingual speaker was used, and the goal was to enable English speakers to speak Spanish and Spanish speakers to speak English. We found that the simple transform was sufficient to convert a voice from one language to the other with a high degree of naturalness. In one case the transformed voice outperformed a native language voice in listening tests. Experiments further indicated that the transform preserved many of the characteristics of the original voice. The degree of accent present can be controlled and naturalness is relatively consistent across a range of accent values.

Scalable Multilingual Frontend for TTS

Alistair Conkie, Andrew Finch

This paper describes progress towards making a Neural Text-to-Speech (TTS) Frontend that works for many languages and can be easily extended to new languages. We take a Machine Translation (MT) inspired approach to constructing the frontend, and model both text normalization and pronunciation on a sentence level by building and using sequence-to-sequence (S2S) models. We experimented with training normalization and pronunciation as separate S2S models and with training a single S2S model combining both functions. For our language-independent approach to pronunciation we do not use a lexicon. Instead all pronunciations, including context-based pronunciations, are captured in the S2S model. We also present a language-independent chunking and splicing technique that allows us to process arbitrary-length sentences. Models for 18 languages were trained and evaluated. Many of the accuracy measurements are above 99%. We also evaluated the models in the context of end-to-end synthesis against our current production system.

Multi-task Learning for Speaker Verification and Voice Trigger Detection

Siddharth Sigtia, Erik Marchi, Sachin Kajarekar, Devang Naik, John Bridle

We describe the design of a voice trigger detection system for smart speakers. In this study, we address two major challenges. The first is that the detectors are deployed in complex acoustic environments with external noise and loud playback by the device itself. Secondly, collecting training examples for a specific keyword or trigger phrase is challenging resulting in a scarcity of trigger phrase specific training data. We describe a two-stage cascaded architecture where a low-power detector is always running and listening for the trigger phrase. If a detection is made at this stage, the candidate audio segment is re-scored by larger, more complex models to verify that the segment contains the trigger phrase. In this study, we focus our attention on the architecture and design of these second-pass detectors. We start by training a general acoustic model that produces phonetic transcriptions given a large labelled training dataset. Next, we collect a much smaller dataset of examples that are challenging for the baseline system. We then use multi-task learning to train a model to simultaneously produce accurate phonetic transcriptions on the larger dataset and discriminate between true and easily confusable examples using the smaller dataset. Our results demonstrate that the proposed model reduces errors by half compared to the baseline in a range of challenging test conditions without requiring extra parameters.

Unsupervised Style and Content Separation by Minimizing Mutual Information for Speech Synthesis

Ting-Yao Hu, Ashish Shrivastava, Oncel Tuzel, Chandra Dhir

We present a method to generate speech from input text and a style vector that is extracted from a reference speech signal in an unsupervised manner, i.e., no style annotation, such as speaker information, is required. Existing unsupervised methods, during training, generate speech by computing style from the corresponding ground truth sample and use a decoder to combine the style vector with the input text. Training the model in such a way leaks content information into the style vector. The decoder can use the leaked content and ignore some of the input text to minimize the reconstruction loss. At inference time, when the reference speech does not match the content input, the output may not contain all of the content of the input text. We refer to this problem as "content leakage", which we address by explicitly estimating and minimizing the mutual information between the style and the content through an adversarial training formulation. We call our method MIST - Mutual Information based Style Content Separation. The main goal of the method is to preserve the input content in the synthesized speech signal, which we measure by the word error rate (WER) and show substantial improvements over state-of-the-art unsupervised speech synthesis methods.

On Modeling ASR Word Confidence

Woojay Jeon, Maxwell Jordan, Mahesh Krishnamoorthy

We present a new method for computing ASR word confidences that effectively mitigates the effect of ASR errors for diverse downstream applications, improves the word error rate of the 1-best result, and allows better comparison of scores across different models. We propose 1) a new method for modeling word confidence using a Heterogeneous Word Confusion Network (HWCN) that addresses some key flaws in conventional Word Confusion Networks, and 2) a new score calibration method for facilitating direct comparison of scores from different models. Using a bidirectional lattice recurrent neural network to compute the confidence scores of each word in the HWCN, we show that the word sequence with the best overall confidence is more accurate than the default 1-best result of the recognizer, and that the calibration method can substantially improve the reliability of recognizer combination.

SNDCNN: Self-Normalizing Deep CNNs With Scaled Exponential Linear Units For Speech Recognition

Zhen Huang, Tim Ng, Leo Liu, Henry Mason, Xiaodan Zhuang, Daben Liu

Very deep CNNs achieve state-of-the-art results in both computer vision and speech recognition, but are difficult to train. The most popular way to train very deep CNNs is to use shortcut connections (SC) together with batch normalization (BN). Inspired by Self- Normalizing Neural Networks, we propose the self-normalizing deep CNN (SNDCNN) based acoustic model topology, by removing the SC/BN and replacing the typical RELU activations with scaled exponential linear unit (SELU) in ResNet-50. SELU activations make the network self-normalizing and remove the need for both shortcut connections and batch normalization. Compared to ResNet- 50, we can achieve the same or lower (up to 4.5% relative) word error rate (WER) while boosting both training and inference speed by 60%-80%. We also explore other model inference optimization schemes to further reduce latency for production use

Leveraging GANs to Improve Continuous Path Keyboard Input Models

Akash Mehra, Jerome R. Bellegarda, Ojas Bapat, Partha Lal, Xin Wang

Continuous path keyboard input has higher inherent ambiguity than standard tapping, because the path trace may exhibit not only local overshoots/undershoots (as in tapping) but also, depending on the user, substantial mid-path excursions. Deploying a robust solution thus requires a large amount of high-quality training data, which is difficult to collect/annotate. In this work, we address this challenge by using GANs to augment our training corpus with user-realistic synthetic data. Experiments show that, even though GAN-generated data does not capture all the characteristics of real user data, it still provides a substantial boost in accuracy at a 5:1 GAN-to-real ratio. GANs therefore inject more robustness in the model through greatly increased word coverage and path diversity.

Embedded Large-Scale Handwritten Chinese Character Recognition

Youssouf Chherawala, Hans Dolfing, Ryan Dixon, Jerome Bellegarda

As handwriting input becomes more prevalent, the large symbol inventory required to support Chinese handwriting recognition poses unique challenges. This paper describes how the Apple deep learning recognition system can accurately handle up to 30,000 Chinese characters while running in real-time across a range of mobile devices. To achieve acceptable accuracy, we paid particular attention to data collection conditions, representativeness of writing styles, and training regimen. We found that, with proper care, even larger inventories are within reach. Our experiments show that accuracy only degrades slowly as the inventory increases, as long as we use training data of sufficient quality and in sufficient quantity.

Related readings and updates.

Lattice-based Improvements for Voice Triggering Using Graph Neural Networks

Voice-triggered smart assistants often rely on detection of a trigger-phrase before they start listening for the user request. Mitigation of false triggers is an important aspect of building a privacy-centric non-intrusive smart assistant. In this paper, we address the task of false trigger mitigation (FTM) using a novel approach based on analyzing automatic speech recognition (ASR) lattices using graph neural networks (GNN). The proposed…
See paper details

Multi-Task Learning for Voice Trigger Detection

We describe the design of a voice trigger detection system for smart speakers. In this study, we address two major challenges. The first is that the detectors are deployed in complex acoustic environments with external noise and loud playback by the device itself. Secondly, collecting training examples for a specific keyword or trigger phrase is challenging resulting in a scarcity of trigger phrase specific training data. We describe a two-stage…
See paper details