Voice search is a rapidly developing technology that is finding more and more applications, including in e-commerce. At this year’s Mobile Trends Awards, the Mobile Support Technology category won edrone for its Mobile Voice Search project. What distinguishes this technology? What applications can voice search have in e-commerce? We invite you to interview Tomasz Michalczyk, Product Manager at edrone
Bartłomiej Lewandowski: Voice search engines have been on the market for some time now. What, then, distinguishes the Mobile Voice Search you have created?
Tomasz Michalczyk: When you say voice search, probably, you mean voice assistants such as Alexa, Siri, or Google Assistant. These solutions have been with us for some time and impact our daily lives. It is convenient to operate devices by voice. However, they do not offer the possibility to browse products in online shops.
A Mobile Voice Search is a search engine developed for e-commerce that allows customers to search for products by voice. It was built, among other things, on the basis of the LLM (Large Language Model), so using it resembles a ‘direct conversation’ with the shop. You can implement it in virtually any e-commerce site and thus facilitate the shopping experience for the customer using a mobile device.
BL: What exactly does “e-commerce dedicated” mean? Why is voice search better for e-commerce than standard search?
TM: The Mobile Voice Search Widget is a solution that works well for e-commerce because our algorithms have learned to understand product features (categories, parameters, etc.) at a much higher level of detail than Google or Siri. In addition, from a customer’s statement worded somewhat arbitrarily, our solution can guess what product features are involved, such as color or size.
From the beginning of the project, we knew that we were creating a tool specifically for use in e-commerce, and this specialization is our advantage, unlike Google or Siri, which touch on a number of other areas.
BL: Can you talk about what technologies you used in your project?
MT: In our project, we used technology from different areas, such as Speech-to-Text, known as ASR Automatic Speech Recognition, for processing and understanding user speech. NLU (Natural Language understanding) to understand user queries, and NLG (Natural Language Generation) to generate prompts.
Machine learning algorithms to improve system performance and, of course, databases and cloud computing to store data and host applications. We are currently working on integrating ChatGPT into our solution.
BL: What was the biggest challenge during the design work?
TM: We had a lot of challenges like that. For me personally, it was coming up with a way to explain to the user how to use our product. The Mobile Voice Search is a new solution, so shoppers weren’t familiar with a similar interface. The intuitiveness of use alone is not always enough in such situations, and additional educational activities are needed.
BL: What are the prospects for the voice search sector’s development? Do you see an opportunity for further growth in this area?
TM: Yes, definitely yes. We are already seeing how much of an impact a well-configured search engine can have on a business. Mobile has an advantage over desktops in terms of e-commerce users every year. It will also become more and more voice-oriented.
It is noticeable that more and more people prefer to record themselves instead of traditional texting – according to data shared by Whatsapp, users sent seven billion voice messages, per day in 2022. There’s more, the world tech giants like Google and Apple are developing their technologies towards voice. As we know, trends from overseas are reaching us very quickly, so we can expect a similar scenario in Poland too.