---
license: apache-2.0
tags:
  - openvoiceos
  - benchmark
  - predictions
  - voice-activity-detection
  - vad
pretty_name: OVOS vad bench — speech-vs-nonspeech
configs:
- config_name: default
  data_files:
  - split: en
    path: predictions/en/*.jsonl
---

# OVOS `vad` bench — `speech-vs-nonspeech`

Per-clip speech / non-speech decisions predictions of the registered
[OVOS Plugin Arena](https://github.com/OpenVoiceOS/ovos-plugin-arena)
`vad` fighters over
[`TigreGotico/not-wake-words-speech-en`](https://huggingface.co/datasets/TigreGotico/not-wake-words-speech-en).

One dedicated repo per modality; one dataset split per language; one JSONL
file per fighter under `predictions/<lang>/<competitor_id>.jsonl`. Rows follow
the arena §3.2 contract (pinned `dataset_revision`, `plugin_version`,
`latency_ms`). Produced by the reproducible benchmark script in the arena repo;
the arena's `assemble` workflow turns these rows into benchmark boards, blind
battle pools and a benchmark-seeded ELO ladder.

Funded by the [NGI0 Commons Fund](https://nlnet.nl/project/OpenVoiceOS) /
[NLnet](https://nlnet.nl) under grant agreement No
[101135429](https://cordis.europa.eu/project/id/101135429), through the
European Commission's [Next Generation Internet](https://ngi.eu) programme.