Search

Recent advances in voice synthesis, coupled with the ease with which speech can be harvested for millions of people, introduce new threats to applications that are enabled by devices such as voice assistants (e.g., Amazon Alexa, Google Home etc.). We explore if unrelated and limited amount of speech from a target can be used to synthesize commands for a voice assistant like Amazon Alexa. More specifically, we investigate attacks on voice assistants with synthetic commands when they match command sources to authorized users, and applications (e.g., Alexa Skills) process commands only when their source is an authorized user with a chosen confidence level. We demonstrate that even simple concatenative speech synthesis can be used by an attacker to command voice assistants to perform sensitive operations. We also show that such attacks, when launched by exploiting compromised devices in the vicinity of voice assistants, can have relatively small host and network footprint. Our results demonstrate the need for better defenses against synthetic malicious commands that could target voice assistants.

Building on the success of large language models (LLMs), recent advancements such as GPT-4o have enabled real-time speech interactions through LLM-based voice assistants, offering a significantly improved user experience compared to traditional text-based interactions. However, the absence of benchmarks designed to evaluate these speech interaction capabilities has hindered progress of LLM-based voice assistants development. Current evaluations focus primarily on automatic speech recognition (ASR) or general knowledge evaluation with clean speeches, neglecting the more intricate, real-world scenarios that involve diverse speaker characteristics, environmental and content factors. To address this, we introduce VoiceBench, the first benchmark designed to provide a multi-faceted evaluation of LLM-based voice assistants. VoiceBench also includes both real and synthetic spoken instructions that incorporate the above three key real-world variations. Extensive experiments reveal the limitations of current LLM-based voice assistant models and offer valuable insights for future research and development in this field., Comment: Work in progress. Data is available at https://github.com/MatthewCYM/VoiceBench

Many companies, including Google, Amazon, and Apple, offer voice assistants as a convenient solution for answering general voice queries and accessing their services. These voice assistants have gained popularity and can be easily accessed through various smart devices such as smartphones, smart speakers, smartwatches, and an increasing array of other devices. However, this convenience comes with potential privacy risks. For instance, while companies vaguely mention in their privacy policies that they may use voice interactions for user profiling, it remains unclear to what extent this profiling occurs and whether voice interactions pose greater privacy risks compared to other interaction modalities. In this paper, we conduct 1171 experiments involving a total of 24530 queries with different personas and interaction modalities over the course of 20 months to characterize how the three most popular voice assistants profile their users. We analyze factors such as the labels assigned to users, their accuracy, the time taken to assign these labels, differences between voice and web interactions, and the effectiveness of profiling remediation tools offered by each voice assistant. Our findings reveal that profiling can happen without interaction, can be incorrect and inconsistent at times, may take several days to weeks for changes to occur, and can be influenced by the interaction modality.

Voice Assistants (VAs) are popular for simple tasks, but users are often hesitant to use them for complex activities like online shopping. We explored whether the vocal characteristics like the VA's vocal tone, can make VAs perceived as more attractive and trustworthy to users for complex tasks. Our findings show that the tone of the VA voice significantly impacts its perceived attractiveness and trustworthiness. Participants in our experiment were more likely to be attracted to VAs with positive or neutral tones and ultimately trusted the VAs they found more attractive. We conclude that VA's perceived trustworthiness can be enhanced through thoughtful voice design, incorporating a variety of vocal tones., Comment: Extended Abstract

Supporting older adults in health self-management is crucial for promoting independent aging, particularly given the growing strain on healthcare systems. While voice assistants (VAs) hold the potential to support aging in place, they often lack tailored assistance and present usability challenges. We addressed these issues through a five-stage design process with older adults to develop a personal health assistant. Starting with in-home interviews (N=17), we identified two primary challenges in older adult's health self-management: health awareness and medical adherence. To address these challenges, we developed a high-fidelity LLM-powered VA prototype to debrief doctor's visit notes and generate tailored medication reminders. We refined our prototype with feedback from co-design workshops (N=10) and validated its usability through in-home studies (N=5). Our work highlights key design features for personal health assistants and provides broader insights into desirable VA characteristics, including personalization, adapting to user context, and respect for user autonomy.

Recent works demonstrate that voice assistants do not perform equally well for everyone, but research on demographic robustness of speech technologies is still scarce. This is mainly due to the rarity of large datasets with controlled demographic tags. This paper introduces the Sonos Voice Control Bias Assessment Dataset, an open dataset composed of voice assistant requests for North American English in the music domain (1,038 speakers, 166 hours, 170k audio samples, with 9,040 unique labelled transcripts) with a controlled demographic diversity (gender, age, dialectal region and ethnicity). We also release a statistical demographic bias assessment methodology, at the univariate and multivariate levels, tailored to this specific use case and leveraging spoken language understanding metrics rather than transcription accuracy, which we believe is a better proxy for user experience. To demonstrate the capabilities of this dataset and statistical method to detect demographic bias, we consider a pair of state-of-the-art Automatic Speech Recognition and Spoken Language Understanding models. Results show statistically significant differences in performance across age, dialectal region and ethnicity. Multivariate tests are crucial to shed light on mixed effects between dialectal region, gender and age.

While standalone Voice Assistants (VAs) are promising to support older adults' daily routine and wellbeing management, onboarding and setting up these devices can be challenging. Although some older adults choose to seek assistance from technicians and adult children, easy set up processes that facilitate independent use are still critical, especially for those who do not have access to external resources. We aim to understand the older adults' experience while setting up commercially available voice-only and voice-first screen-based VAs. Rooted in participants observations and semi-structured interviews, we designed a within-subject study with 10 older adults using Amazon Echo Dot and Echo Show. We identified the values of the built-in touchscreen and the instruction documents, as well as the impact of form factors, and outline important directions to support older adult independence with VAs., Comment: 5 pages, 1 figure, 1 table, Companion Publication of the 2023 ACM Designing Interactive Systems Conference, July 2023, Pages 164-168

Voice assistants (VAs) are becoming a feature of our everyday life. Yet, the user experience (UX) is often limited, leading to underuse, disengagement, and abandonment. Co-designing interactions for VAs with potential end-users can be useful. Crowdsourcing this process online and anonymously may add value. However, most work has been done in the English-speaking West on dialogue data sets. We must be sensitive to cultural differences in language, social interactions, and attitudes towards technology. Our aims were to explore the value of co-designing VAs in the non-Western context of Japan and demonstrate the necessity of cultural sensitivity. We conducted an online elicitation study (N = 135) where Americans (n = 64) and Japanese people (n = 71) imagined dialogues (N = 282) and activities (N = 73) with future VAs. We discuss the implications for coimagining interactions with future VAs, offer design guidelines for the Japanese and English-speaking US contexts, and suggest opportunities for cultural plurality in VA design and scholarship., Comment: 21 pages

Our work addresses the challenges older adults face with commercial Voice Assistants (VAs), notably in conversation breakdowns and error handling. Traditional methods of collecting user experiences-usage logs and post-hoc interviews-do not fully capture the intricacies of older adults' interactions with VAs, particularly regarding their reactions to errors. To bridge this gap, we equipped 15 older adults' homes with smart speakers integrated with custom audio recorders to collect "in-the-wild" audio interaction data for detailed error analysis. Recognizing the conversational limitations of current VAs, our study also explored the capabilities of Large Language Models (LLMs) to handle natural and imperfect text for improving VAs. Midway through our study, we deployed ChatGPT-powered VA to investigate its efficacy for older adults. Our research suggests leveraging vocal and verbal responses combined with LLMs' contextual capabilities for enhanced error prevention and management in VAs, while proposing design considerations to align VA capabilities with older adults' expectations.

In-car Voice Assistants (VAs) play an increasingly critical role in automotive user interface design. However, existing VAs primarily perform simple 'query-answer' tasks, limiting their ability to sustain drivers' long-term attention. In this study, we investigate the effectiveness of an in-car Robot Assistant (RA) that offers functionalities beyond voice interaction. We aim to answer the question: How does the presence of a social robot impact user experience in real driving scenarios? Our study begins with a user survey to understand perspectives on in-car VAs and their influence on driving experiences. We then conduct non-driving and on-road experiments with selected participants to assess user experiences with an RA. Additionally, we conduct subjective ratings to evaluate user perceptions of the RA's personality, which is crucial for robot design. We also explore potential concerns regarding ethical risks. Finally, we provide a comprehensive discussion and recommendations for the future development of in-car RAs., Comment: Submitted to ACM Transactions on Computer-Human Interaction

The adoption of voice assistants like Alexa or Siri has grown rapidly, allowing users to instantly access information via voice search. Query suggestion is a standard feature of screen-based search experiences, allowing users to explore additional topics. However, this is not trivial to implement in voice-based settings. To enable this, we tackle the novel task of suggesting questions with compact and natural voice hints to allow users to ask follow-up questions. We define the task, ground it in syntactic theory and outline linguistic desiderata for spoken hints. We propose baselines and an approach using sequence-to-sequence Transformers to generate spoken hints from a list of questions. Using a new dataset of 6681 input questions and human written hints, we evaluated the models with automatic metrics and human evaluation. Results show that a naive approach of concatenating suggested questions creates poor voice hints. Our approach, which applies a linguistically-motivated pretraining task was strongly preferred by humans for producing the most natural hints., Comment: Accepted as Long Paper at EMNLP'23 Findings

In the context of a voice assistant system, steering refers to the phenomenon in which a user issues a follow-up command attempting to direct or clarify a previous turn. We propose STEER, a steering detection model that predicts whether a follow-up turn is a user's attempt to steer the previous command. Constructing a training dataset for steering use cases poses challenges due to the cold-start problem. To overcome this, we developed heuristic rules to sample opt-in usage data, approximating positive and negative samples without any annotation. Our experimental results show promising performance in identifying steering intent, with over 95% accuracy on our sampled data. Moreover, STEER, in conjunction with our sampling strategy, aligns effectively with real-world steering scenarios, as evidenced by its strong zero-shot performance on a human-graded evaluation set. In addition to relying solely on user transcripts as input, we introduce STEER+, an enhanced version of the model. STEER+ utilizes a semantic parse tree to provide more context on out-of-vocabulary words, such as named entities that often occur at the sentence boundary. This further improves model performance, reducing error rate in domains where entities frequently appear, such as messaging. Lastly, we present a data analysis that highlights the improvement in user experience when voice assistants support steering use cases., Comment: EMNLP 2023 Industry Track

Commercial voice assistants are largely feminized and associated with stereotypically feminine traits such as warmth and submissiveness. As these assistants continue to be adopted for everyday uses, it is imperative to understand how the portrayed gender shapes the voice assistant's ability to mitigate errors, which are still common in voice interactions. We report a study (N=40) that examined the effects of voice gender (feminine, ambiguous, masculine), error mitigation strategies (apology, compensation) and participant's gender on people's interaction behavior and perceptions of the assistant. Our results show that AI assistants that apologized appeared warmer than those offered compensation. Moreover, male participants preferred apologetic feminine assistants over apologetic masculine ones. Furthermore, male participants interrupted AI assistants regardless of perceived gender more frequently than female participants when errors occurred. Our results suggest that the perceived gender of a voice assistant biases user behavior, especially for male users, and that an ambiguous voice has the potential to reduce biases associated with gender-specific traits.

Spoken dialogue systems have transformed human-machine interaction by providing real-time responses to queries. However, misunderstandings between the user and system persist. This study explores the significance of interactional language in dialogue repair between virtual assistants and users by analyzing interactions with Google Assistant and Siri, focusing on their utilization and response to the other-initiated repair strategy "huh?" prevalent in human-human interaction. Findings reveal several assistant-generated strategies but an inability to replicate human-like repair strategies such as "huh?". English and Spanish user acceptability surveys show differences in users' repair strategy preferences and assistant usage, with both similarities and disparities among the two surveyed languages. These results shed light on inequalities between interactional language in human-human interaction and human-machine interaction, underscoring the need for further research on the impact of interactional language in human-machine interaction in English and beyond., Comment: In WTF Workshop Proceedings (arXiv:2401.04108) held in conjunction with the ACM conference on Conversational User Interfaces (CUI), 19 - 21/07 2023, in Eindhoven, The Netherlands

Joint intent detection and slot filling, which is also termed as joint NLU (Natural Language Understanding) is invaluable for smart voice assistants. Recent advancements in this area have been heavily focusing on improving accuracy using various techniques. Explainability is undoubtedly an important aspect for deep learning-based models including joint NLU models. Without explainability, their decisions are opaque to the outside world and hence, have tendency to lack user trust. Therefore to bridge this gap, we transform the full joint NLU model to be `inherently' explainable at granular levels without compromising on accuracy. Further, as we enable the full joint NLU model explainable, we show that our extension can be successfully used in other general classification tasks. We demonstrate this using sentiment analysis and named entity recognition., Comment: Accepted at CIKM 2023

Conventional Voice Assistants (VAs) rely on traditional language models to discern user intent and respond to their queries, leading to interactions that often lack a broader contextual understanding, an area in which Large Language Models (LLMs) excel. However, current LLMs are largely designed for text-based interactions, thus making it unclear how user interactions will evolve if their modality is changed to voice. In this work, we investigate whether LLMs can enrich VA interactions via an exploratory study with participants (N=20) using a ChatGPT-powered VA for three scenarios (medical self-diagnosis, creative planning, and discussion) with varied constraints, stakes, and objectivity. We observe that LLM-powered VA elicits richer interaction patterns that vary across tasks, showing its versatility. Notably, LLMs absorb the majority of VA intent recognition failures. We additionally discuss the potential of harnessing LLMs for more resilient and fluid user-VA interactions and provide design guidelines for tailoring LLMs for voice assistance.

Little research has investigated the design of conversational styles of voice assistants (VA) for adults in their later adulthood with varying personalities. In this Wizard of Oz experiment, 34 middle-aged (50 to 64 years old) and 24 older adults (65 to 80 years old) participated in a user study at a simulated home, interacting with a VA using either formal or informal language. Older adults with higher agreeableness perceived VA as being more likeable than middle-aged adults. Middle-aged adults showed similar technology acceptance toward the informal and formal VA, and older adults preferred using informal VA, especially those with low agreeableness. Further, while both middle-aged and older adults frequently anthropomorphized VAs by using human metaphors for them, older adults compared formal VA with professionals (e.g., librarians, teachers) and informal VA with their close ones (e.g., spouses, relatives). Overall, the conversational style showed differential effects on the perceptions of middle-aged and older adults, suggesting personalized design implications.

Language speakers often use what are known as repair initiators to mend fundamental disconnects that occur between them during verbal communication. Previous research in this field has mainly focused on the human-to-human use of repair initiator. We proposed an examination of dialogue repair structure wherein the dialogue initiator is human and the party that initiates or responds to the repair is a virtual assistant. This study examined the use of repair initiators in both English and Spanish with two popular assistants, Google Assistant and Apple's Siri. Our aim was to codify the differences, if any, in responses by voice assistants to dialogues in need of repair as compared to human-human dialogues also in need of repair. Ultimately the data demonstrated that not only were there differences between human-assistant and human-human dialogue repair strategies, but that there were likewise differences among the assistants and the languages studied., Comment: 2021, non-print, unpublished version

Voice assistants help users make phone calls, send messages, create events, navigate, and do a lot more. However, assistants have limited capacity to understand their users' context. In this work, we aim to take a step in this direction. Our work dives into a new experience for users to refer to phone numbers, addresses, email addresses, URLs, and dates on their phone screens. Our focus lies in reference understanding, which becomes particularly interesting when multiple similar texts are present on screen, similar to visual grounding. We collect a dataset and propose a lightweight general-purpose model for this novel experience. Due to the high cost of consuming pixels directly, our system is designed to rely on the extracted text from the UI. Our model is modular, thus offering flexibility, improved interpretability, and efficient runtime memory utilization., Comment: 7 pages, Accepted to ACL Industry Track 2023

Streaming Automatic Speech Recognition (ASR) in voice assistants can utilize prefetching to partially hide the latency of response generation. Prefetching involves passing a preliminary ASR hypothesis to downstream systems in order to prefetch and cache a response. If the final ASR hypothesis after endpoint detection matches the preliminary one, the cached response can be delivered to the user, thus saving latency. In this paper, we extend this idea by introducing predictive automatic speech recognition, where we predict the full utterance from a partially observed utterance, and prefetch the response based on the predicted utterance. We introduce two personalization approaches and investigate the tradeoff between potential latency gains from successful predictions and the cost increase from failed predictions. We evaluate our methods on an internal voice assistant dataset as well as the public SLURP dataset., Comment: Accepted for Interspeech 2023

The adoption of multimodal interactions by Voice Assistants (VAs) is growing rapidly to enhance human-computer interactions. Smartwatches have now incorporated trigger-less methods of invoking VAs, such as Raise To Speak (RTS), where the user raises their watch and speaks to VAs without an explicit trigger. Current state-of-the-art RTS systems rely on heuristics and engineered Finite State Machines to fuse gesture and audio data for multimodal decision-making. However, these methods have limitations, including limited adaptability, scalability, and induced human biases. In this work, we propose a neural network based audio-gesture multimodal fusion system that (1) Better understands temporal correlation between audio and gesture data, leading to precise invocations (2) Generalizes to a wide range of environments and scenarios (3) Is lightweight and deployable on low-power devices, such as smartwatches, with quick launch times (4) Improves productivity in asset development processes.