codesparktech/vlotspraak-dutch-asr

Vlotspraak Dutch ASR

Proprietary Dutch automatic speech recognition by CodeSpark Tech B.V. (product: VeraDossier). On the public Dutch benchmarks, scored with the benchmarks' own official tooling, it currently posts the lowest word error rate we are aware of for Dutch — including a #1 on the Dutch community Common Voice benchmark and a lead over NVIDIA Canary-1b-v2 and Parakeet-TDT-0.6b-v3 on the HuggingFace Open ASR harness.

Every number below is scored with the venue's own scorer and travels with committed, inspectable output. Weights are commercial; the evaluation is fully open.

Full receipts, contamination check and reproduction: https://github.com/CodeSpark-Tech/dutch-asr-benchmarks

Common Voice (nl) — official ASR_NL / NIST sclite

On the Dutch community ASR benchmark (Common Voice 17.0 NL test), scored with the official ASR_NL_benchmark Docker tool (NIST sclite):

System	WER	Test set
Vlotspraak Dutch ASR	3.5%	full 11,266 clips
faster-whisper v3 (previous best on the board)	4.3%	full 11,266 clips
our faster-whisper-large-v3 control (same pipeline)	4.3%	identical clips

The control reproduces the board's incumbent exactly, so the improvement is not a decoding artifact. A train-on-test check (Common Voice filenames are stable across releases) found only 9 of 11,266 test clips (0.08%) in our training data; the full and leak-free numbers differ by 0.1pp. Submitted for inclusion: ASR_NL_results PR #7.

HuggingFace Open ASR harness (nl) — vs the multilingual leaders

Same harness, same scorer (normalizer.eval_utils.score_results), eval-set identity verified:

Benchmark	Vlotspraak	NVIDIA Canary-1b-v2	NVIDIA Parakeet-TDT-0.6b-v3
FLEURS NL	4.63	5.27	6.50
Common Voice NL	4.30	5.78	5.47
MLS NL	5.11	9.64	11.27

WER (%), lower is better. Our Parakeet run reproduces NVIDIA's published FLEURS figure to ~0.2pp, confirming the setup. Proposed for the board's Dutch track: open_asr_leaderboard #187. Note: Vlotspraak is a 1.55B Whisper-large-v3 derivative and is slower (lower RTFx) than the 0.6B NeMo models — we optimise for accuracy, not throughput.

Held-out medical (Dutch)

On the Radboud/HoMed Medicijnjournaal corpus (sealed held-out; official ASR_NL NIST-sclite): 11.4% WER — below the trained Wav2vec2.0 (12.8%) and Kaldi-NL (16.1%) baselines, with no medical training data. Indicative on 30 of 35 episodes (disclosed); being closed with the corpus authors.

Intended use

Dutch transcription for regulated, on-premise / air-gapped deployment (healthcare, GGZ, government, finance). Runs entirely inside the customer perimeter — no external cloud dependency. Data is protected with hybrid post-quantum encryption (NIST-standardised ML-KEM-1024 + ML-DSA-65, FIPS 203/204, over AES-256-GCM).

Availability

Weights are commercial and licensed per-tenant (Vlotspraak Commercial License), not downloadable here. The full evaluation is open: per-sample sclite output, the harness, the contamination check, and a one-command reproduction — contact info@codesparktech.nl. This card exists so the results are public and independently checkable. Building Dutch healthcare software and want to license or validate the engine on your own test data under NDA? Same contact.

Provenance & honesty

• Fine-tune of the open yuriyvnv/whisper-large-v3-high-mixed-nl (Apache-2.0) on rights-clean public Dutch data (Common Voice 25, FLEURS, VoxPopuli, MLS). Not trained from scratch; not open-weight. Upstream attribution preserved per Apache-2.0.
• Common Voice: fine-tuned on the CV train split; the board's incumbent is zero-shot. Fine-tuning on a benchmark's train split is legitimate (that is its purpose) but is a real advantage on Common Voice's recording conditions — stated plainly, as the board's existing XLS-R FT entry also is.
• Medical figure is on 30/35 HoMed episodes (disclosed).
• Numbers scored on different harnesses use different normalizers and splits and are not interchangeable; each table above names its scorer.

Citation

Technical note in preparation (arXiv link to be added).

© CodeSpark Tech B.V., Amsterdam. Model IP is proprietary; upstream components carry their own licenses.