Amazon's Alexa has been able to recognize different speakers in a household based on their voice for a couple of years now, and now Alexa Skill developers too will be able to make the most of the feature.
Alexa devices often have more than one speaker interacting with it, and different speakers in turn are linked to different accounts in third-party apps like Uber, for example. Until now, the assistant could only differentiate between speakers if they manually switched to their personal Amazon account.
So when it came to asking the assistant to perform custom skills, like shuffle play your Spotify playlist or predict your ETA to the office, you would first have had to make sure the assistant was switched to your personal account settings.
SEE: Amazon Alexa: An insider's guide (free PDF)
The company said: "Now, your skill can determine who is speaking when customers engage naturally with your skill, no longer needing to switch between Alexa accounts to access individual preferences".
Users can configure a voice profile by ordering Alexa to "learn my voice". Then, when the assistant recognizes a speaker, it will generate and send a personID – an anonymous identifier made of characters and numbers – to the skill, so that it responds appropriately based on who is speaking.
"Personalizing your skill experience can range from a friendly greeting to responses based on a customer's likes, dislikes, interests, or account history," said Amazon.
The company named several third-party applications that speaker recognition will be compatible with. This includes Uber, meaning that Alexa will be able to bill the right account for a ride without users having to previously log into it.
Another compatible application is 7-Minute Workout. This will let the assistant immediately deliver the speaker's personalized fitness routine, history and goals.
The company mentioned other apps such as Excite Horoscope, Vodafone, Japanese game app Jyanken, and said it was working on others, including bus commuting app OneBusAway.
Customers will be able to opt out of the feature even if they have already created a voice profile; although Amazon said that the personID it generates for each voice profile does not contain any personally identifiable information.
At the same time, it warned that skill personalization won't be available for child-directed applications, or skills that require sensitive user information such as data related to health insurance.
Voice recognition is not an entirely new feature for Alexa. Two years ago, the Echo, Echo Dot and Echo Show started recognizing different voice profiles, but until now the technology was limited to calling, messaging, flash briefing, shopping and the Amazon Music Unlimited Family Plan.
The initial launch coincided with Google's development of Voice Match, which lets users link up to six voices with speakers fitted with Google Assistant.
Personalizing skills is only the latest iteration of Amazon's voice recognition technology. It has already been open to a select number of developers as a trial feature for almost two months.
Amazon provided a guide to build a sample skill with personalization via GitHub.