Is talking to an automated teller machine natural and fun?

Area of Science:

• Human-Computer Interaction
• Speech Technology
• Banking Technology

Background:

• Traditional automated teller machines (ATMs) rely on manual interfaces, presenting potential usability and affective challenges.
• Automatic Speech Recognition (ASR) technology offers a novel interaction method for ATMs, with potential benefits for user experience.

Purpose of the Study:

• To investigate the usability and affective aspects of interacting with an ATM using ASR technology.
• To design and evaluate a speech-driven ATM interface, accommodating bilingual users.
• To compare the user experience of a hybrid speech ATM with a traditional manual ATM.

Main Methods:

• Two experiments were conducted: dialogue pattern analysis for interface design and a Wizard-of-Oz simulation for a bilingual speech ATM.
• A hybrid speech ATM was evaluated against a simulated manual ATM for usability and affective factors.
• ANOVA was used to analyze usability, affective data, task completion, and transaction errors.

Main Results:

• The speech-driven ATM successfully accommodated bilingual users (Bahasa Melayu and English).
• Significant differences were found in task completion and transaction errors between the hybrid speech ATM and the manual ATM.
• The hybrid speech system provided a more pleasurable interaction, with users expressing willingness to use speech ATMs.

Conclusions:

• Despite current limitations in ASR technology, speech-enabled ATMs can enhance user experience.
• The hybrid speech ATM demonstrates potential for more natural and enjoyable banking interactions.
• Further development of ASR for ATMs is warranted for public adoption.