苹果端侧语音识别反超Whisper
推荐指数 76.0 NO. 009 · 2026.07.14
发布2026/07/13Score264Comments122
为什么值得看
苹果新推出的SpeechAnalyzer API在LibriSpeech基准测试中WER全面优于Whisper全系模型,速度是Whisper Small的3倍且无需额外模型体积。对需要在iOS/macOS上做实时语音交互的开发者,这是目前最优的端侧方案。
编辑判断
WhisperKit团队此前花了大量精力做CoreML优化,结果苹果自家API直接碾压,这暴露了第三方在苹果生态做底层优化的天花板。更关键的是SpeechAnalyzer作为系统API,意味着苹果可能把语音能力下沉为平台级基础设施,类似当年CoreML对第三方ML框架的挤压。
做语音AI应用的团队需要重新评估技术栈:如果目标用户主要在苹果生态,继续自研Whisper部署的ROI在下降。但跨平台产品仍需保留Whisper方案,SpeechAnalyzer的iOS/macOS 26限制和安卓缺失决定了它无法成为唯一选项。另外注意WER差距主要来自噪声场景,clean speech下各方案差距缩小,如果产品场景以安静环境为主,迁移紧迫性没那么高。
社区反馈
意见分歧 117 条评论
核心争论:苹果端侧语音识别是否真最优,还是Parakeet等第三方模型更实用
相关内容
mlx-examples部署指南:将Whisper语音识别部署到iOS设备 介绍通过Apple MLX框架将Whisper模型部署到iOS设备的方法,解决移动端语音识别性能瓶颈,实现高效低延迟的端侧语音转文字。 Whisper Notes App:離線語音轉文字 基于OpenAI Whisper Large V3 Turbo的iOS/Mac离线语音转文字应用,零云端上传,医疗法律场景适用,已有10,000+用户。 Whisper Explained:端到端语音识别新范式深度解读 深度解析Whisper架构、算法与源码,讨论large-v3、turbo、distil-whisper等模型变体的部署选型策略。
Author here. I ship both Apple speech engines plus WhisperKit side by side in a transcription app, which made it possible to run all five through identical production code on the same audio: LibriSpeech test-clean and test-other, 5,559 utterances, fully on-device on an M2 Pro. Apple published no acc
At this point, I would not recommend ignoring Parakeet TDT 0.6b v2/v3 (english-only versus multilingual). Those models have been out for a year, give or take, and they are both accurate and fast. I would choose Parakeet over Whisper in almost all situations these days. Parakeet works great even
Why use relatively ancient models like whisper and not e. g. parakeet?