model stringclasses 3
values | params stringclasses 3
values | arch stringclasses 2
values | wer_clean_pct float64 1.42 1.49 | wer_other_pct float64 4.73 6.26 | rtfx int64 53 451 | vram_gb float64 2 4.6 | dataset stringclasses 1
value | utts_per_split int64 200 200 |
|---|---|---|---|---|---|---|---|---|
parakeet-tdt-0.6b-v2 | 0.6B | TDT transducer | 1.49 | 4.73 | 451 | 2 | LibriSpeech | 200 |
whisper-large-v3 | 1.55B | attention enc-dec | 1.47 | 5.96 | 53 | 4.6 | LibriSpeech | 200 |
whisper-large-v3-turbo | 809M | attention enc-dec | 1.42 | 6.26 | 117 | 2.5 | LibriSpeech | 200 |
Sovereign ASR Bench — RTX 5090
Local, self-hosted automatic speech recognition benchmarks on one RTX 5090 32GB.
Part of the WITCHEER local-AI rig. Methodology that matters: load-once measurement (so
RTFx times transcription, not model load), one shared text normalizer applied to every
model output and reference, and micro-averaged WER (total errors / total reference
words — the LibriSpeech standard). The board lives as data in board.csv (shown in the viewer).
Board — LibriSpeech (test-clean / test-other), 200 utts/split, load-once
| model | params | arch | WER clean | WER other | RTFx | VRAM |
|---|---|---|---|---|---|---|
| parakeet-tdt-0.6b-v2 | 0.6B | TDT transducer | 1.49% | 4.73% | 451× | 2.0 GB |
| whisper-large-v3 | 1.55B | attention enc-dec | 1.47% | 5.96% | 53× | 4.6 GB |
| whisper-large-v3-turbo | 809M | attention enc-dec | 1.42% | 6.26% | 117× | 2.5 GB |
Clean read speech is a three-way tie (saturated). On the noisy split the 0.6B Parakeet
transducer wins on WER and runs 4–9× faster on the least VRAM — the smallest model is
the most noise-robust. Full writeup + mechanism in asr-head-to-head.md.
Method
- Runner (load-once):
scripts/asr_bench.py— parakeet.cppbench --manifest(TDT) + whisper.cpp multi-file, both on sm_120 CUDA. - Scorer (dep-free):
scripts/asr_report.py— micro-WER + RTFx + peak VRAM.
Caveats
200 utts/split is directional (ranking solid, exact margin provisional vs the full 2,939-utt test-other). English-only. Number-words-vs-digits are not reconciled — counted as errors equally for every model.
- Downloads last month
- 22