Open Audio Judge GitHub repository
Audio model evaluation

Two focused leaderboards, one evaluation toolkit

Explore speech recognition and speech generation separately, then use the open-source repository for methodology, prompts, and reproducible workflows.

Verified benchmark

ASR Leaderboard

Compare transcription models by semantic accuracy, workload strengths, shared failures, and case-level evidence.

3 MLX models35 clips105 evaluations
Open the ASR leaderboard
Illustrative preview

TTS Leaderboard

Explore the TTS evaluation workflow, sample diagnostics, model/category comparisons, and report interface.

440-case eval library3-model previewNot a verified ranking
Open the TTS leaderboard

Take this workflow to another benchmark

Browse the Audio Benchmark Index, choose a compatible speech benchmark, and reuse Open Audio Judge's case-to-report workflow with small task-specific changes to the case adapter and scoring rubric.

Want the methodology or code?

Visit the Open Audio Judge repository for judge prompts, CLI workflows, evaluation cases, and report generation.