AudioBench Results

Evaluation reports for speech / audio model experiments.

Reports

• Canary — Luciole 1B — dataset v2 — first experiments
  2026-04-17
• Canary — Qwen — dataset v1 — quick training
  2025-12-31