GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
Playedwell03
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
Playedwell03
Model Tree
Qwen/Qwen3-ASR-1.7B-hf
Input Modalities
Output Modalities
Supported Functionality
Held-out test, 500 utterances not seen during training or checkpoint selection, normalized CER (whitespace/punctuation stripped), corpus-level (total edits / total reference characters):
| Condition | Base model | This adapter |
|---|---|---|
| Clean | 0.298 | 0.161 |
| Siren, 5dB SNR | 0.315 | 0.184 |
| Background noise, 10dB | 0.338 | 0.217 |
| Crowd babble, 10dB | 0.368 | 0.245 |
| Background noise, 0dB | 0.512 | 0.344 |
| Crowd babble, 5dB | 0.502 | 0.336 |
| Crowd babble, 0dB (noise = speech loudness) | 0.741 | 0.597 |
The clean-condition improvement is statistically significant (bootstrap 95% CI excludes 0). At 0dB crowd babble the model is still frequently wrong — this remains an open problem, not solved by this adapter.
Given silence, background noise, or a siren alone (no speech), the base model hallucinates text (e.g. "그렇죠.") 14–100% of the time depending on the noise type; this adapter outputs nothing in that situation (0% across all three tested noise-only conditions, n=50 each).
Qwen/Qwen3-ASR-1.7B-hf, bf16q_proj,k_proj,v_proj,o_proj,gate_proj,up_proj,down_proj) — by suffix matching this also touches
the audio encoder's q/k/v projections (but not its out_proj/fc1/fc2, which use different names)accelerate)20 hours from AI Hub's 119 emergency-call speech dataset (Korean, citizen-to-119-dispatcher calls), balanced across 4 disaster categories at 5h each, utterances with personal-information masking tags removed from training. Train/validation splits follow AI Hub's own official split (verified: zero call-ID overlap between them). Noise is synthetic, not real recordings.
system prompt.
Passing a vocabulary hint via the system role at inference degrades CER (tested: 0.182→0.269 on a
related checkpoint, 95% CI excludes 0). Do not use this feature with this adapter without retraining.python
import torchfrom peft import PeftModelfrom transformers import AutoProcessor, Qwen3ASRForConditionalGeneration BASE = "Qwen/Qwen3-ASR-1.7B-hf"ADAPTER = "Playedwell03/qwen3-asr-1.7b-119ko-v5" processor = AutoProcessor.from_pretrained(BASE)model = Qwen3ASRForConditionalGeneration.from_pretrained(BASE, dtype=torch.bfloat16, device_map="cuda")model = PeftModel.from_pretrained(model, ADAPTER)model.eval() inputs = processor.apply_transcription_request(audio="call.wav", language="Korean").to(model.device, model.dtype)output_ids = model.generate(**inputs, max_new_tokens=256)generated = output_ids[:, inputs["input_ids"].shape[1]:]print(processor.decode(generated, return_format="transcription_only")[0])Requires transformers>=5.13.0 and peft.
import torchfrom peft import PeftModelfrom transformers import AutoProcessor, Qwen3ASRForConditionalGeneration BASE = "Qwen/Qwen3-ASR-1.7B-hf"ADAPTER = "Playedwell03/qwen3-asr-1.7b-119ko-v5" processor = AutoProcessor.from_pretrained(BASE)model = Qwen3ASRForConditionalGeneration.from_pretrained(BASE, dtype=torch.bfloat16, device_map="cuda")model = PeftModel.from_pretrained(model, ADAPTER)model.eval() inputs = processor.apply_transcription_request(audio="call.wav", language="Korean").to(model.device, model.dtype)output_ids = model.generate(**inputs, max_new_tokens=256)generated = output_ids[:, inputs["input_ids"].shape[1]:]print(processor.decode(generated, return_format="transcription_only")[0])