With One Voice: Composing a Travel Voice Assistant from Re-purposed Models

TL;DR

This paper demonstrates how a travel voice assistant can be efficiently built by re-purposing existing machine learning models, reducing development time and costs while maintaining functional performance.

Contribution

It provides a detailed comparison of custom-built versus re-purposed models for a travel voice assistant, highlighting practical trade-offs and implementation insights.

Findings

Re-purposed models can achieve comparable performance to custom solutions.

Using existing models significantly reduces development time and costs.

Data-driven trade-offs inform effective implementation choices.

Abstract

Voice assistants provide users a new way of interacting with digital products, allowing them to retrieve information and complete tasks with an increased sense of control and flexibility. Such products are comprised of several machine learning models, like Speech-to-Text transcription, Named Entity Recognition and Resolution, and Text Classification. Building a voice assistant from scratch takes the prolonged efforts of several teams constructing numerous models and orchestrating between components. Alternatives such as using third-party vendors or re-purposing existing models may be considered to shorten time-to-market and development costs. However, each option has its benefits and drawbacks. We present key insights from building a voice search assistant for Booking.com search and recommendation system. Our paper compares the achieved performance and development efforts in dedicated tailor-made solutions against existing re-purposed models. We share and discuss our data-driven decisions about implementation trade-offs and their estimated outcomes in hindsight, showing that a fully functional machine learning product can be built from existing models.