Voice Lab

Product Design

COMPANY

Replica Studios

PRODUCT

Blending voices in Voice Lab

ROLE

Lead Product Designer ✦ Product Owner

YEAR

2024 – 2025

Replica Studios had already earned the trust of leading game studios, filmmakers, and immersive storytellers for its high-fidelity AI voice technology. But while the platform offered a growing library of premium voices, one creative limitation kept surfacing:


“If I want to be able to create a voice that sounds like my character,
not just pick one that’s close?”


Voice Lab was our answer to that question. Built for creators, not machine learning engineers, it offered a way to design entirely new AI voices that are custom-tailored to fit a character’s personality, narrative role, and emotional arc.

Prompt for Voice Lab

Blending Voices in Voice Lab

MY ROLE

As Lead Product Designer, I was responsible for the end-to-end design of Voice Lab, from early concept, wireframes, prototypes, to production launch, working closely with the CEO, CTO, machine learning, and engineering teams.


Also, as a co-product owner, I helped establish the roadmap and create detailed product tickets for the engineering team with User Stories, Requirements, Acceptance Criteria, Designs, Prototypes, etc.


My contributions also included:


  • Designing an intuitive tool balancing creative freedom and technical complexity

  • Using research and customer insights to shape the product vision and priorities

  • Supporting non-technical users with helpful suggestions, previews, and guidance

  • Considering ethical voice design principles in every step of the design process

  • Doing testing and quality assurance on the product

  • Producing art and marketing assets for Voice Lab

DESIGN APPROACH

Where existing tools were technical and rigid, we envisioned Voice Lab as a creative playground, a place where writers, animators, and designers could experiment, iterate, and truly own their characters’ voices.


  1. Prompt-based customization
    We introduced natural language input as a creative onramp. Users could describe the voice they wanted, for example, “a stoic space captain with a hidden softness,” and receive AI-generated voices that matched their intent.


  1. Voice blending and tuning
    Creators used curated presets to guide the performer’s accent and speech cadence, while also controlling the tone and timbre of the voice through a flexible voice blend. These references acted as “vocal mood boards,” helping fine-tune the voice’s style and delivery with precision.


  1. Real-time iteration
    We built in fast previewing and side-by-side comparison tools so users could audition multiple voice variants before selecting their final match, cutting iteration time from days to minutes.


  1. Ethical by design
    All voices were trained using licensed data from professional actors, with full commercial rights and revenue share built in. We treated ethical scaffolding not as an afterthought, but as a design pillar.

Custom Voice Blends

Voice Lab example prompts

Adjusting performance pitch + pace

Adjusting voice blend weights

OUTCOME + REFLECTION

Voice Lab unlocked a new level of creative control for studios and storytellers alike.


  • Accelerated character development, reducing voice prototyping time from weeks to hours

  • Enabled bespoke voice creation, moving beyond library selection into expressive, purpose-built AI voices

  • Supported responsible innovation, balancing creative freedom with voice actor protection and transparent usage rights


What started as a power-user feature quickly became a strategic differentiator, strengthening Replica’s position as not just a voice provider but a full-stack creative solution.


Designers, writers, and actors could shape voice performances together in a transparent, expressive space.