OPEN BENCHMARK FOR CHARACTER VOICE DESIGN

Measuring the art of character voice design

A high-quality benchmark of 1,000 character voice-design tasks spanning a broad range of media genres and real-world creative use cases.

View Leaderboard View complete benchmark on Hugging Face

MULTI-MODEL RESULTS

Character Voice Design Leaderboard

RankCompanyModel

BreezeBlueBreeze TTS 2

78.0270898.5%

XiaomiMiMo-V2.5-TTS

72.7820296.3%

InworldInworld-VD

64.6137898.2%

AlibabaFun-AudioGen-VD

63.8213698.6%

ElevenLabsEleven v3

58.8811199.7%

AlibabaQwen3-TTS-VD

58.488399.6%

OpenBMBVoxCPM2

55.2850998.5%

METRICS

Three measures of effective character voice design.

We evaluate role fit, voice diversity, and script pass—capturing both individual performance and range across the full cast.

Run benchmark on GitHub

Role Fit

Whether the generated voice belongs to the character — age, texture, temperament, and delivery rated 1–5 against the full role profile, reported out of 100. Clips that fail the transcript gate are excluded.

Voice Diversity

The number of acoustically distinct voice clusters across the generated cast — whether it still sounds like different people, or collapses into a few house voices. Complements Role Fit: a model can follow single prompts well while reusing the same few voices.

Transcript Pass

Whether each clip clearly speaks its intended line, judged separately from voice-design quality. Minor pronunciation or clarity issues pass; noticeable substitutions, omissions, or insertions rate as degraded; rewritten or skipped lines fail. The reported score is the share of clips rated Pass.

DATASET

1,000 voices across seven worlds of storytelling

Curated from CharacterCodex, each task pairs an anonymized character prompt with a natural preview script—testing distinctive voice creation without relying on known names or franchises.