←Back to NewsAI News/AudiomodelAudioGpusMoondream Shrinks Parakeet Redux Speech Recognition to 178 MBMoondream shrank NVIDIA's Parakeet speech model from 1.2GB to 178MB using ternary weights, hitting 113x realtime on CPU while matching accuracy.SourceAlphaSignalPublishedSep 25, 2026, 12:58 PMAuthorAlphaSignal NewsroomRead1 min readMoondream shrank NVIDIA's Parakeet speech model from 1.2GB to 178MB using ternary weights, hitting 113x realtime on CPU while matching accuracy.Reporting is indexed from AlphaSignal. Rights remain with the original publisher and cited sources.Read original report ↗Next readsSarvam · newsSarvam Releases Saaras V4 to Transcribe All 22 Indian Languages at OnceAlphaSignal · modelEdge0 Ships Audio8 ASR Infinite to Transcribe Speech for Hours StraightPerplexity · newsPerplexity Brings Portable Computer to AMD Ryzen AI Max PCs
