Unverified50% confidenceOpinionExact time
Whistle's transcription quality is close to Whisper Base and Moonshine Tiny V2.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/21/2026
First Seen
Valid until: 1/19/2027
Sources
17MB Speech Transcription Model Whistle: A New Option for On-Device Deployment
youtubeNeuralNine10/21/2026
Related Entities
Related Claims
UnverifiedThe MiniFreak 49's core sound source (sound engine, oscillator types, and filters) is almost identical to the current MiniFreak.62% similarUnverifiedWhisper is released as open source under the MIT license, covering multiple sizes ranging from Tiny (39M parameters) to Large (1.5B parameters).61% similarUnverifiedWhen running the Whisper model on-device, transcription accuracy and speed are affected by hardware performance, and older devices may deliver a poor experience.60% similarUnverified在 M1 芯片上运行 Whisper base 模型的转录速度可达音频时长的 10-15 倍，即 1 分钟音频仅需 4-6 秒完成转录60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1082569API
curl https://kongchang.com/api/v1/knowledge/claims/1082569MCP
get_claim(id=1082569)