SPA: Web-based Platform for easy Access to Speech Processing Modules

SPA: Web-based Platform for easy Access to Speech Processing Modules Fernando Batista1,2, Pedro Curto1, Isabel Trancoso1,3, Alberto Abad1,3, Jaime Ferreira1, Eug´enio Ribeiro1,3, Helena Moniz1,4, David M. de Matos1,3, Ricardo Ribeiro1,2 1 L2F – Spoken Language Systems Laboratory, INESC-ID Lisboa 2 ISCTE-IUL - Instituto Universit´ario de Lisboa, Lisboa, Portugal 3 Instituto Superior T´ecnico, Universidade de Lisboa, Portugal 4 FLUL/CLUL, Universidade de Lisboa, Portugal fernando.batista@inesc-id.pt, pkurto@gmail.com, isabel.trancoso@inesc-id.pt Abstract This paper presents SPA, a web-based Speech Analytics platform that integrates several speech processing modules and that makes it possible to use them through the web. It was developed with the aim of facilitating the usage of the modules, without the need to know about software dependencies and specific configurations. Apart from being accessed by a web-browser, the platform also provides a REST API for easy integration with other applications. The platform is flexible, scalable, provides authentication for access restrictions, and was developed taking into consideration the time and effort of providing new services. The platform is still being improved, but it already integrates a considerable number of audio and text processing modules, including: Automatic transcription, speech disfluency classification, emotion detection, dialog act recognition, age and gender classification, non-nativeness detection, hyper- articulation detection, dialog act recognition, and two external modules for feature extraction and DTMF detection. This paper describes the SPA architecture, presents the already integrated modules, and provides a detailed description for the ones most recently integrated. Keywords: Web interface, Web services, Speech modules, Speech Analytics, REST API 1. Introduction World Wide Web is becoming a development plat- form (Cervinschi et al., 2010) and web-based applications are emerging everyday. Web-based applications are fairly easy to develop, use, and maintain, thus being a way of pro- viding access to existing and new resources for the whole community. This paper presents a web-based Speech Ana- lytics platform – SPA – that makes it possible to use exist- ing speech processing modules through the web. The original motivation for the SPA platform was to an- swer the increasing number of requests received for tran- scribing audio/video files in European Portuguese. It was firstly created to provide a simple interface, easy to use by non-expert users, but requests have multiplied and diversi- fied, namely in terms of languages (Spanish, English) and varieties covered (European, Brazilian), domains (broad- cast news, interviews), etc.. On the other hand, although the majority of the SPA users are only interested in the au- tomatic transcripts, other technology partners showed in- terest in obtaining information about characteristics of the speakers (e.g. male/female) and all the other metadata that could be retrieved from the speech signal (e.g. language, recording conditions, etc.). In fact, a vast number of other audio and text processing modules have been developed at the Spoken Language Lab (L2F), including: speech disflu- ency classification, emotion detection, dialog act recogni- tion, age and gender classification, amongst others. Thus, a major goal was to develop a platform that could integrate most the existing modules as services, ease the integra- tion of new services, …

Content truncated to 500 words