“…Notable exceptions include the Verbmobil project (Wahlster, 2000); designers identified many concerns that influenced system design, including recovery from error. Similarly, the BBN TransTalk device (Prasad et al, 2013), developed as part of the military-funded TRANSTAC research, also included a way for at least one party to estimate the quality of the translation. These systems are designed from a single perspective, and reflect the power dynamics in the environment (Risku et al, 2021;Paullada, 2020).…”
Machine translation models are embedded in larger user-facing systems. Although model evaluation has matured, evaluation at the systems level is still lacking. We review literature from both the translation studies and HCI communities about who uses machine translation and for what purposes. We emphasize an important difference in evaluating machine translation models versus the physical and cultural systems in which they are embedded. We then propose opportunities for improved measurement of user-facing translation systems. We pay particular attention to the need for design and evaluation to aid engendering trust and enhancing user agency in future machine translation systems.
“…Notable exceptions include the Verbmobil project (Wahlster, 2000); designers identified many concerns that influenced system design, including recovery from error. Similarly, the BBN TransTalk device (Prasad et al, 2013), developed as part of the military-funded TRANSTAC research, also included a way for at least one party to estimate the quality of the translation. These systems are designed from a single perspective, and reflect the power dynamics in the environment (Risku et al, 2021;Paullada, 2020).…”
Machine translation models are embedded in larger user-facing systems. Although model evaluation has matured, evaluation at the systems level is still lacking. We review literature from both the translation studies and HCI communities about who uses machine translation and for what purposes. We emphasize an important difference in evaluating machine translation models versus the physical and cultural systems in which they are embedded. We then propose opportunities for improved measurement of user-facing translation systems. We pay particular attention to the need for design and evaluation to aid engendering trust and enhancing user agency in future machine translation systems.
“…To our knowledge, there are no conversational translation systems where the system is a fluent party in the conversation. Some systems situate speech-to-speech translation in humanoid robotic form [43] and others provide structured prompts to the original speaker to disambiguate speech [36], but none emulate the clarifying questions of a professional interpreter. We propose that the app can employ a variety of strategies to improve dyadic conversation by exploiting awareness of model quality and semantics.…”
Translation apps and devices are often presented in the context of providing assistance while traveling abroad. However, the spectrum of needs for cross-language communication is much wider. To investigate these needs, we conducted three studies with populations spanning socioeconomic status and geographic regions: (1) United States-based travelers, (2) migrant workers in India, and (3) immigrant populations in the United States. We compare frequent travelers' perception and actual translation needs with those of the two migrant communities. The latter two, with low language proficiency, have the greatest translation needs to navigate their daily lives. However, current mobile translation apps do not meet these needs. Our findings provide new insights on the usage practices and limitations of mobile translation tools. Finally, we propose design implications to help apps better serve these unmet needs.
“…Previously, speech recognition on handheld computers and smartphones has been studied in the DARPA sponsored Transtac Program, where speech-to-speech translation systems were developed on the phone [3,4,5]. In the Transtac systems, Gaussian mixture models (GMMs) were used to as acoustic models.…”
In this paper we describe the development of an accurate, smallfootprint, large vocabulary speech recognizer for mobile devices. To achieve the best recognition accuracy, state-of-the-art deep neural networks (DNNs) are adopted as acoustic models. A variety of speedup techniques for DNN score computation are used to enable real-time operation on mobile devices. To reduce the memory and disk usage, on-the-fly language model (LM) rescoring is performed with a compressed n-gram LM. We were able to build an accurate and compact system that runs well below real-time on a Nexus 4 Android phone.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.