zai-org/GLM-ASR-Nano-2512
Open-source speech recognition model (~2B) with strong dialect support (Cantonese and others) and robust low-volume speech transcription
Outperforms Whisper V3 on multiple benchmarks at compact 1.5B active / 2B total size
Open-source speech recognition model (~2B) with strong dialect support (Cantonese and others) and robust low-volume speech transcription
Outperforms Whisper V3 on multiple benchmarks at compact 1.5B active / 2B total size
View on HuggingFace
View on ModelScope