Unverified50% confidenceFactExact time
端侧模型的典型代表包括 Whisper 系列轻量版本（tiny/base/small）、Meta 的 MMS（Massively Multilingual Speech）以及各厂商自研的压缩模型
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/21/2026
First Seen
Valid until: 11/19/2026
Sources
Related Claims
UnverifiedThe Whisper model is available in multiple sizes ranging from tiny to large and can run on local devices.80% similarUnverifiedWhisper 提供从 tiny 到 large 的多个规格版本，在准确率与计算资源消耗之间提供不同权衡73% similarUnverified2022年OpenAI发布的Whisper证明了单一神经网络可以直接将音频映射为文字，绕开手工设计声学模型的传统路径72% similarUnverifiedThe workflow offers two voice generation modes: a lip-sync mode (which freezes the user's voice into the audio latent space) and a model-generated speech mode.71% similarUnverified标准Whisper模型在处理歌声时准确率会下降，社区开发了whisper-timestamped等变体实现逐词级别时间轴对齐71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/782813API
curl https://kongchang.com/api/v1/knowledge/claims/782813MCP
get_claim(id=782813)