As a concept, speech recognition technology has been around for quite a while. Popularized in science fiction like 2001: A Space Odyssey, The Hitchhiker’s Guide to the Galaxy, and Star Trek (“Tea. Earl Grey. Hot.”), it’s only been in the last few years that speech recognition software has become refined enough for everyday use.
One of the most common uses of voice technology is speaking a query into a search engine. Using voice to search has been growing steadily in recent years; one in five adults use voice search on their cell phone once a month, and 76% use their smart speakers to search at least once a week. Siri, Alexa, and Cortana are all names your users know and interact with every day. These bots largely rely on searches behind the scenes to make sense of what humans are saying.
The exciting thing about the advancement of voice search though is that it’s now easy for anyone to have this feature on their own website.
At Sajari we believe that interacting with all forms of data can be made much easier. Here, I’ll be focusing on how we offer voice search to our customers at Sajari, why it is useful, and whether or not you should consider a voice search strategy.
The magic behind voice-to-text technology
As more people began speaking queries instead of typing them, we realized that we would need to offer similar functionality to our site and ecommerce search customers. We use Chrome’s in-browser voice detection, which runs off Google’s Speech to Text.
We send a voice signal to Google, which transcribes it into text and sends it back to Sajari. We use this as a query to search within a customer’s index and return results - all in mere milliseconds. It’s a similar process to a regular search operation, but instead of a query being typed, it’s spoken first.
It's possible in the future we'll use an API to send voice signals. MuleSoft gives a great overview of APIs and how they’re used to send and receive data.
The actual process of how Google turns voice into text is interesting. Using machine learning, algorithms have been trained to associate typed words with voice patterns, and over time the machine learns the relationship between how a word sounds, and how it is written as text.
We then combine voice detection with other Sajari features to make it more useful. By looking at the context around a query, Sajari can understand natural language and can extract rich information from keywords to filter results and make them more relevant.
For example, a user could speak the query “basketball games near me, next week”. Sajari can extract keywords from this voice command and apply time and geolocation filters. We will treat “basketball” as the subject, “games” as an event, “near me” as a location and “next week” as a timeframe.
So when a user speaks this query, Sajari will automatically apply filters and return results for basketball games, located near the user, from the next coming week. Pretty advanced for a voice query, right?
This functionality used to be restricted to voice assistants or web search engines, but using Sajari, smart and intuitive voice search can now be used for individual sites.
The UX benefits of voice search
A site search controlled by one’s voice is particularly useful for those who find it hard to search in the “traditional” way, where a user clicks the search bar, types in a series of characters, and hits return or clicks “search”.
For people with certain disabilities, the elderly, or even those that can’t type as speedily or accurately as others (for example, those with RSI, carpal tunnel, or even a broken wrist or arm) searching using their voice is a gamechanger. Speaking a query out loud speeds up a process that could take minutes instead of seconds.
It’s like having a direct connection to a search engine - queries can be spoken and results returned instantly, with minimal physical obstacles for the user.
Voice controlled devices are extremely useful for those who have trouble typing.
Voice search is also useful when the user is busy doing something else and doesn’t want to interrupt this by searching.
For example, a user could be searching on your site for an error code from their television. Instead of memorizing a few characters, typing these into a search bar, looking up again, typing in a few more and pressing return, the user just needs to speak the error code out loud from their TV.
Catering to users who might be multitasking is important when considering user experience. Sometimes users can be impeded with a traditional search field, and it’s important to make navigating and searching a site as painless as possible.
The third reason why voice search could be useful, and also the most simple to understand, is that it’s….cool. Using the power of voice to complete a task is impressive, and having the same functionality is an option that users love to see on sites.
It’s incredibly easy to offer your users the same functionality when they search your site. All a user needs to do is select the voice search icon next to a search bar and speak their query. Browsers will need access to the device’s microphone, but once this is selected for the first time future permissions are usually not required for the same domain.
Requirements for a voice search strategy
If you’re evaluating whether your site needs a voice search strategy, it’s worth looking at the points above and deciding whether it will help your users.
If your site has a lot of users, and you know that some of those users have trouble typing, then it’s definitely worth implementing voice search functionality.
Opening up the accessibility for search is necessary as more and more data is uploaded online. More people will need to search sites to find this data, so it’s important to accommodate the increasing number of users that can’t access traditional search.
Similarly, if you want to give your users an easier way to search while multi-tasking then a voice search strategy could prove useful. One of our customers uses voice search as many of their queries are for historical law cases, and they didn’t want their users typing long search queries found in physical books.
If you’re looking for a way to dazzle your users and keep up with current search trends, then you should be looking to implement a voice search strategy. Voice powered devices and software are becoming the norm for users, and it’s a great way to ensure they find what they’re looking for as quickly as possible.