Nelvox Accurate STT — Dutch / Flemish
General beta release channel
Nelvox Accurate STT is an accuracy-oriented automatic speech recognition product for Dutch and Flemish speech.
This repository is the canonical beta distribution channel for the generally offered Dutch/Flemish Accurate STT product. Its stable product identity does not depend on one particular model architecture, checkpoint or decoding implementation.
⚠️ Proprietary — use requires a written Agreement. This model is licensed under the Proprietary Software License in
LICENSE. Downloading or possessing it does not grant any right to use it: use is permitted only under a separate signed Agreement with Nelvox. See nelvox.com.
Access
Access requests are reviewed manually; submitting a request does not guarantee approval. Repository approval, downloading the files, or possessing a copy does not itself grant usage rights — use requires an applicable written Agreement with Nelvox. Authorized participants should request access with the Hugging Face account communicated to Nelvox.
Model overview
| Field | Public information |
|---|---|
| Product | Nelvox Accurate STT |
| Task | Automatic speech recognition / speech-to-text |
| Language | Dutch / Flemish (nl) |
| Channel | Beta |
| Access | Manual review by Nelvox |
Exact release, build, artifact, runtime, inference-profile and checksum
information is provided to approved users in RELEASE.md and nelvox-model.json.
Intended use
The model is intended for authorized evaluation and integration of Dutch/Flemish speech-to-text in accordance with the applicable Agreement. It is not a certified verbatim transcription system, and human review is required wherever transcription errors could have material consequences.
Releases and reproducibility
main may move when Nelvox promotes a new general release. Reproducible
evaluations and deployments must pin the immutable version tag stated in
RELEASE.md. Only promoted, generally distributable Accurate STT releases belong
in this repository; internal experiments and customer-specific adaptations remain
in separate repositories.
Limitations
Quality may vary with dialect, accent, speaking style, vocabulary, recording conditions, background noise, overlapping speakers and degraded audio. The model may omit, substitute, repeat or incorrectly generate words. Approved users should validate every release on representative audio before adopting it.
For approved users
Installation and usage instructions are in USAGE.md. Release record, integrity
(checksums) and machine-readable metadata are in RELEASE.md,
nelvox-model.json, checksums.sha256 and CHANGELOG.md — all included in the
repository and available once access has been granted.
License
Proprietary. See LICENSE. No use without a written Agreement with
Nelvox — nelvox.com.
Contact
Visit nelvox.com for general product enquiries or authorized-access questions.