IBM 发布 Granite Speech 5.0 Turbo CTC:超快语音识别模型
原标题:Extremely Fast and Accurate Transcription with Granite Speech 5.0 Turbo CTC
AI 摘要
IBM 发布 Granite Speech 5.0 Turbo CTC 系列,包含两个 470M 参数的英语语音识别模型,在 NVIDIA H200 GPU 上实现超过 12,600 RTFx 的吞吐量,可每秒转录超过 3.5 小时的语音。模型采用编码器-only 设计,相比前代速度提升 20 倍以上,适合边缘设备使用。非商业版本(CC-BY-NC-SA-4.0)在 OpenASR 上 WER 为 4.85%,Apache 2.0 版本为 5.00%。
正文节选
Today we're delighted to announce the release of two new models in the Granite Speech family: compact, 470M-parameter English speech recognition models that pair strong accuracy with unprecedented speed — over 12,600 RTFx on an NVIDIA H200 GPU, meaning they can transcribe more than 3.5 hours of speech in one second using batched inference. To get a sense of the models' responsiveness, check out our WebGPU demo of streaming speech recognition. Note that the demo only runs on Chrome or Edge browse