跳到正文
HNHacker News·
暂不在当前实时榜单

Show HN: Nari Qwen3-TTS and Qwen3-ASR – High accuracy, low latency and cost

AI 摘要

Nari's Qwen3-ASR Fast model achieved a #1 ranking in Time-to-Final-Segment (TTFS) with a p50 of 44 ms. Its Word Error Rate (WER) was 3.6%, placing it #2 behind AssemblyAI’s Universal 3.5 Pro, which had a WER of 3.5%. These benchmark values were based on Coval’s 1-day view as of September 14, 2026, at 15:00 UTC, with WER pooled across datasets. The rankings exclude dedicated inference endpoints.

为什么是这条

This report highlights Nari's Qwen3-ASR Fast model achieving the #1 ranking in Time-to-Final-Segment, unlike other models that may prioritize only Word Error Rate.

时间与来源

时间显示为 UTC

显示时区:UTC

本地时区尚不可用,暂时显示 UTC。

收录当时偏移:UTC+02026年9月14日 18:00 UTC

收录
2026年9月14日 18:00
来源类型
未分类

本站未收录正文。

前往源站阅读 →
来源·Hacker News·narilabs.com