Qwen3 ASR 1.7B: transcribe.cpp GGUF GGUF conversions of Qwen/Qwen3 ASR 1.7B for use with transcribe.cpp. Ported from upstream commit 7278e1e70fe206f11671096ffdd38061171dd6e5, pinned 2026 04 19. Validated against the qwen asr 0.0.6 reference at transcribe.cpp commit 3f61df7 on 2026 04 20. Offline multilingual speech to text. Same audio LLM architecture as the 0.6B variant (bidirectional audio encoder feeding a Qwen3 causal LM with audio token injection), wider: encoder d model=1024 (16 heads), LM hidden size=2048 , intermediate size=6144 . Auto detects the audio's language across 30 languages and emits the transcript in that language. Takes a 16 kHz mono WAV; explicit language hints are not supported at this time. Downloads Quantization Download Size WER (LibriSpeech test clean) : : BF16 Qwen3 ASR 1.7B BF16.gguf 3.80 GB 1.62% F16 Qwen3 ASR 1.7B F16.gguf 3.81 GB 1.62% Q8 0 Qwen3 ASR 1.7B Q8 0.gguf 2.04 GB 1.61% Q6 K Qwen3 ASR 1.7B Q6 K.gguf 1.58 GB 1.65% Q5 K M Qwen3 ASR 1.7B Q5 K M.gguf 1.41 GB 1.65% Q4 K M Qwen3 ASR 1.7B Q4 K M.gguf 1.23 GB 1.81% WER measured on the full LibriSpeech test clean split (2620 English utterances) with the Whisper style English text normalizer and jiwer…
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy