Six voices, not sixty: how we chose the narrators for Viola

Every voice provider offers dozens of narrators, and the easy product decision is to expose all of them in a long list. We shipped Viola with six. People sometimes ask why so few. The honest answer is that six is how many we could make good.

A long list is a decision pushed onto the user

Sixty voices means a person has to audition voices instead of listening to their scene. That is work, and it happens at exactly the moment they wanted to relax. A short list of distinct, well-tuned voices lets them choose in seconds and get on with the point.

What we listened for

• Pace. Slow enough for the last minutes of the day, with natural pauses. Most default settings are tuned for audiobooks or assistants and are too quick.
• Warmth. A voice that sounds like it is on your side. Hard to define, instantly recognisable.
• Consistency. The same voice reading a hopeful scene and a quiet one should feel like the same person in two moods, not two people.
• Clarity at low volume. Viola is often played quietly. Some voices lose their consonants when quiet; those were cut.

How we tested

The same scene, read by every candidate, listened to with eyes closed in a quiet room, at the volume people actually use before sleep. Not in a meeting, not on laptop speakers. We kept the ones that made us want to keep listening and cut the ones that made us want to open our eyes.

Variety that matters

The six cover a range of tone rather than a range of demographics for its own sake: some lighter, some deeper, some brisker, some slower. People pick by feel, and the set gives them real choices without a catalogue.

What we would add

A voice only earns its place when it is clearly different from the six and tuned as carefully. We would rather add one a year that people love than ten that people scroll past.