Back to blog
BLOG

NeuroVox May 2026 Update: New Languages, Improved Commands & Sub-Second Latency

NeuroVox2026-05-205 min read

The May 2026 NeuroVox update is live. On the menu: 6 new languages, latency reduced to under one second on major language pairs, a new !speed command, multi-pair mode, and improved accuracy on regional accents. Here's everything that's changing.

6 new languages supported. NeuroVox goes from 24 to 30 supported languages with the addition of Dutch (nl), Polish (pl), Hungarian (hu), Czech (cs), Romanian (ro), and Vietnamese (vi). These languages are accessible via the !lang command — for example !lang nl for Dutch. Speech recognition and voice synthesis quality matches the bot's established languages, powered by OpenAI's Whisper Large v3 model.

Latency reduced to under 1 second. The translation pipeline has been optimized end-to-end. On the most common language pairs (fr↔en, fr↔es, en↔es, en↔ja, fr↔de), median latency is now down to 0.8 seconds. Speech feels nearly instant in conversation. On less common languages, latency stays under 1.5 seconds. This improvement is especially noticeable in competitive gaming where every half-second matters.

New !speed command. You can now adjust the speed of voice synthesis with !speed [0.75 / 1.0 / 1.25 / 1.5]. The default remains 1.0. Some communities prefer faster speech to avoid interrupting conversation flow (1.25 or 1.5); others prefer slower reading for language learners (0.75). The setting is saved per server and persists until changed.

Multi-pair mode. Until now, NeuroVox handled a single active language pair per voice channel at a time. With multi-pair mode (available on Pro and Server plans), multiple people speaking different languages in the same voice channel are all translated simultaneously. One member speaks French, another Spanish, a third Japanese — each is translated for the others in real time. NeuroVox automatically detects each speaker's language and routes the translation to the right recipients.

Improved accuracy on regional accents. The underlying Whisper model has been fine-tuned on regional accent data: Quebec, Belgian, Swiss, Brazilian, Mexican, and several regional Spanish accents. Transcription accuracy on these language variants improves by 8–15% compared to the previous version.

How to update? NeuroVox updates automatically — no action required on your end. If you already have the bot on your server, the new features are immediately available. If you haven't tried NeuroVox yet, head to neuro-vox.online to add it to your Discord server in 30 seconds.

NeuroVox May 2026 Update

The May 2026 update brings 6 new languages, latency under 1 second, the new !speed command for voice synthesis, and the multi-pair mode for multilingual voice channels.

Try NeuroVox Free