The AI arena is free today

Open Superagent
All benchmarks

Pronunciation Robustness

PaperImplementation

Progress Over Time

Interactive timeline showing model performance evolution on Pronunciation Robustness

State-of-the-art frontier
Open
Proprietary

Pronunciation Robustness Leaderboard

1 models
ContextCostLicense
1
Fish Audio
Fish Audio Drama 3
Fish Audio
35B——
Notice missing or incorrect data?

Sub-benchmarks

Pronunciation Robustness (Context)

Artificial Analysis Pronunciation Robustness subcategory: words whose reading depends on context.

audio•Max 1

Pronunciation Robustness (Exact Sequences)

Artificial Analysis Pronunciation Robustness subcategory: exact sequences such as codes, paths, and emails.

audio•Max 1

Pronunciation Robustness (Names)

Artificial Analysis Pronunciation Robustness subcategory: names of brands, places and technical terms.

audio•Max 1

Pronunciation Robustness (Shorthand)

Artificial Analysis Pronunciation Robustness subcategory: shorthand including numbers, dates, units, and notation.

audio•Max 1
About this benchmark

What is Pronunciation Robustness?

Artificial Analysis Pronunciation Robustness evaluates how often text-to-speech models correctly read hard words in English sentences (numbers, codes, brand names, and words whose reading depends on context).

Pronunciation Robustness is a audio benchmark evaluating models on general and audio tasks. LLM Stats tracks 1 models on this benchmark, scored on a 0–1 scale. The current average is 0.9, with the leader at 0.9.

Compare leaders on the best AI for general and best AI for audio leaderboards.

Current leaders

Fish Audio Drama 3 from Fish Audio currently leads the Pronunciation Robustness leaderboard with a score of 0.932 across 1 evaluated AI models.

1Fish Audio Drama 3Fish Audio93.2%

FAQ

Common questions about the Pronunciation Robustness benchmark and leaderboard.

What is the Pronunciation Robustness benchmark?

Artificial Analysis Pronunciation Robustness evaluates how often text-to-speech models correctly read hard words in English sentences (numbers, codes, brand names, and words whose reading depends on context).

What is the Pronunciation Robustness leaderboard?

The Pronunciation Robustness leaderboard ranks 1 AI models based on their performance on this benchmark. Currently, Fish Audio Drama 3 by Fish Audio leads with a score of 0.932. The average score across all models is 0.932.

What is the highest Pronunciation Robustness score?

The highest Pronunciation Robustness score is 0.932, achieved by Fish Audio Drama 3 from Fish Audio.

How many models are evaluated on Pronunciation Robustness?

1 models have been evaluated on the Pronunciation Robustness benchmark, with 0 verified results and 1 self-reported results.

Where can I find the Pronunciation Robustness paper?

The Pronunciation Robustness paper is available at https://artificialanalysis.ai/text-to-speech#pronunciation-robustness. The paper details the methodology, dataset construction, and evaluation criteria.

Where can I find the Pronunciation Robustness dataset?

The Pronunciation Robustness dataset is available at https://artificialanalysis.ai/text-to-speech#pronunciation-robustness.

What categories does Pronunciation Robustness cover?

Pronunciation Robustness is categorized under general and audio. The benchmark evaluates audio models.

Are there variants of Pronunciation Robustness?

Yes. Pronunciation Robustness has 4 related variants: Pronunciation Robustness (Context), Pronunciation Robustness (Exact Sequences), Pronunciation Robustness (Names), Pronunciation Robustness (Shorthand).

How recent are the Pronunciation Robustness leaderboard results?

The Pronunciation Robustness leaderboard was last updated in November 2026 and currently includes 1 evaluated models.