Recognition Performance of Selected Speech Recognition APIs – A Longitudinal Study

Abstract

Within the last five years, the availability and usability of interactive voice assistants have grown. Thereby, the development benefits mostly from the rapidly increased cloud-based speech recognition systems. Furthermore many cloud-based services, such as Google Speech API, IBM Watson, and Wit.ai, can be used for personal applications and transcription tasks. As these tasks vary in their domain, their complexity as well as in their interlocutor, it is challenging to select a suitable cloud-based speech recognition service. As the update-process of online-services can be completely handled in the back-end, client applications do not need to be updated and thus improved accuracies can be expected within certain periods. This paper contributes to the field of automatic speech recognition, by comparing the performance of speech recognition between the above-mentioned cloud-based systems on German samples of high-qualitative spontaneous human-directed and device-directed speech as well as noisy device-directed speech over a period of eight months.