Siri Voice Recognition
Siri Voice Recognition
Bringing Siri to the family
Bringing Siri to the family

COMPANY: APPLE

COMPANY: APPLE

COMPANY: APPLE

CATEGORY: ARTIFICIAL INTELLIGENCE

CATEGORY: ARTIFICIAL INTELLIGENCE

CATEGORY: ARTIFICIAL INTELLIGENCE

Family HomePod
Family HomePod
Family HomePod

In a shared device environment like HomePod, it's important for each family member to have easy access to their own content. Voice identification is a crucial component of this feature. However, implementing it in Apple's ecosystem of products poses immense challenges. How do we ensure access to everyone's content while respecting their privacy? Harry Simmonds, our founder at Studio Elros, led the development of a personalised music experience using voice recognition to address these challenges.

What was done

  • Personalised music experiences for each family member 

  • Interconnecting the family music experience 

  • Re-architecture of Siri’s music search engine

01.

Managing Music

The user flow to ensure a specific user received their desired content required a complete overhaul of how we searched for music via voice. New concepts and use cases became present, such as shared music, appropriate music, and accessing everyone's library, which may contain songs from CDs downloaded years ago. Creating an experience around this required detailed thought, particularly given the risk of confusing the user with the amount of possibilities and unique solutions.

02.

Engineering for Privacy

Storing voice profiles securely is crucial and requires careful handling to avoid the loss of any user identifiers associated with their voice. This is one of the reasons why Apple prioritises on-device data, giving users control over their own data. However, we still need to link a user's voice profile with their corresponding music profile in the cloud.


Designing a system that addresses privacy concerns is a rigorous process, especially when dealing with millions of users. Our approach involved peer architecture reviews, end-to-end testing, and iterative design to create a foolproof solution.

03.

User experience for incorrect voice profiling

Determining the correct course of action when voice profiling fails is not always straightforward. Should the user be prompted to confirm their identity, or should the system select what it believes to be the best music? What is the error rate for voice recognition? Do all users have a subscription, and should the music be personalised or generic?


These factors can complicate the user experience, particularly with the vast array of music choices available. While there may be unique solutions for each situation, the risk of confusing the user is high, especially if they are unaware that voice profiling has failed. By conducting user studies and analysing internal usage data, we can identify patterns and determine the most effective user experience. As voice recognition technology improves, these considerations may change.

Media Credit: Apple Inc.

Let’s talk

Subscribe to our upcoming newsletter for occasional updates! We only send emails when we have something to say.

© 2023 Studio Elros

·

Let’s talk

Subscribe to our upcoming newsletter for occasional updates! We only send emails when we have something to say.

© 2023 Studio Elros

·

Let’s talk

Subscribe to our upcoming newsletter for occasional updates! We only send emails when we have something to say.

© 2023 Studio Elros

·