handy models

speech-to-text benchmarks for transcribe.cpp, the engine behind Handy

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

Finnish · 11 models · left is slower, down is better · log speed axis

AMD Ryzen 4750U GPU (Vulkan)
pareto frontier (no model is both faster and more accurate) other models
1 whisper-large-v3most accurate 7.737.12–8.32 2.1× 0.6× 1.55 GB Apache-2.0
2 whisper-large-v3-turbo 8.297.65–8.93 3.4× 0.8× 0.83 GB Apache-2.0
3 canary-1b-v2!overall pickfast pick 8.868.21–9.55 13.2× 6.7× 1.10 GB CC-BY-4.0
4 parakeet-tdt-0.6b-v3 13.3012.48–14.10 12.5× 7.5× 0.72 GB CC-BY-4.0
5 parakeet-primeline 13.3912.60–14.20 12.5× 7.5× 0.72 GB CC-BY-4.0
6 whisper-medium 14.6713.94–15.57 4.3× 1.1× 0.77 GB Apache-2.0
7 nemotron-3.5-asr-streaming-0.6b 21.9120.93–22.94 14.5× 7.5× 0.70 GB OpenMDW-1.1
8 Qwen3-ASR-1.7B 25.4824.58–26.47 3.8× 2.0× 2.04 GB Apache-2.0
9 whisper-small 26.4825.40–27.59 12.1× 3.4× 0.25 GB Apache-2.0
10 Qwen3-ASR-0.6B 46.4945.20–47.90 8.0× 4.3× 0.79 GB Apache-2.0
11 Fun-ASR-MLT-Nano-2512! 68.1666.48–69.98 9.0× 4.5× 0.83 GB FunASR Model Open Source License Agreement v1.1