best model by language
Benchmarked on the FLEURS test split at Q8_0. Error is WER — lower is better. Speed is × realtime on Handy target hardware.
accuracy vs speed
Indonesian · 11 models
AMD Ryzen 4750U GPU (Vulkan)
pareto frontier (no model is both faster and more accurate) other models
highlighted rows are recommended picks or on pareto frontier
| | | | | | |
| 1 | Qwen3-ASR-1.7Baccurate pick | 5.37 | 4.1× | 2.5× | 2.04 GB | Apache-2.0 |
| 2 | whisper-large-v3 | 6.08 | 2.0× | 1.0× | 1.55 GB | Apache-2.0 |
| 3 | whisper-large-v3-turbo | 7.20 | 3.1× | 1.4× | 0.83 GB | Apache-2.0 |
| 4 | whisper-large-v2 | 7.43 | 2.0× | 1.0× | 1.55 GB | Apache-2.0 |
| 5 | Fun-ASR-MLT-Nano-2512!overall pick | 7.52 | 8.8× | 5.5× | 0.83 GB | FunASR Model Open Source License Agreement v1.1 |
| 6 | Qwen3-ASR-0.6B | 8.49 | 9.7× | 5.3× | 0.79 GB | Apache-2.0 |
| 7 | whisper-large | 8.93 | 2.0× | 1.0× | 1.55 GB | Apache-2.0 |
| 8 | whisper-medium | 10.79 | 4.1× | 1.9× | 0.77 GB | Apache-2.0 |
| 9 | whisper-smallfast pick | 18.02 | 11.7× | 5.9× | 0.25 GB | Apache-2.0 |
| 10 | whisper-base | 38.02 | 29.0× | 19.3× | 0.08 GB | Apache-2.0 |
| 11 | whisper-tiny | 60.49 | 43.1× | 39.8× | 0.04 GB | Apache-2.0 |