Models

Reson8 开源语音识别模型 Rumble-2-Flash,主打嘈杂真实场景

Why we picked it

Reson8 开源了 Rumble-2-Flash,专治重口音、多人抢话和电话噪音这些让 Whisper 头疼的音频。40 多种语言,体积小到手机能跑,权重随便商用。做语音转写产品的开发者可以拿自己的脏数据测一测。

Reson8 发布开源语音识别模型 Rumble-2-Flash,权重采用 Apache 2.0 协议,可免费下载并本地部署。该模型专门针对真实世界音频训练,重点解决重口音、多人重叠说话、电话通话和背景噪声这四类常见失败场景。

Image · elvis
Body

Rumble-2-Flash 支持 40 多种语言,据称在同等场景下表现优于体积大 20 倍的模型,且模型足够小,可以在手机上直接运行。开发者可在 Hugging Face(Reson8/Rumble-2-Flash)等平台获取权重。

对需要处理客服电话、会议录音或多语种语音的企业来说,Apache 2.0 协议意味着可以商用且音频数据不出本地,与 Whisper 等开源方案形成新的选择。

Original · elvis

Speech-to-text still breaks on real-world audio. Rumble-2-Flash from Reson8 is a free, open-weights speech-to-text model trained on audio that usually trips up these systems. Think heavy accents, people talking over each other, phone calls, and background noise. It covers 40+ languages. The weights are Apache 2.0, so you can run it on your own machine and keep your audio local. huggingface.co/Reson8/Rumble-… RR @raoulritterr Rumble-2-flash is open weights you can get it in all your favorite places. It’s better than models 20x it’s size small enough to run on your phone HF link below

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Models" desk. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include Rumble-2-Flash, Resonance8, speech recognition, open-source models.

Source attribution: curated from elvis (@omarsar0).