GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
BuzzASR
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Run this model inference with full control and performance in your environment.
Model Details
Model Provider
BuzzASR
Model Tree
openai/whisper-large-v3
Input Modalities
Output Modalities
Supported Functionality
| Test set | CER | WER | Whisper-large-v3 (zero-shot) CER |
|---|---|---|---|
| FLEURS | 4.51 | 14.6 | 8.74 |
| Common Voice 25 | 16.97 | 20.42 | 26.36 |
| Combined | 12.69 | 18.33 | 22.9 |
~1.8x CER reduction over Whisper zero-shot on the combined test set.
python
import torch, torchaudiofrom transformers import WhisperForConditionalGeneration, WhisperProcessormodel = WhisperForConditionalGeneration.from_pretrained("BuzzASR/swahili", torch_dtype=torch.float16).to("cuda").eval()proc = WhisperProcessor.from_pretrained("BuzzASR/swahili")wav, sr = torchaudio.load("audio.wav") # 16 kHz monofeats = proc(wav[0], sampling_rate=16000, return_tensors="pt").input_features.to("cuda").half()ids = model.generate(feats, num_beams=1, no_repeat_ngram_size=3, repetition_penalty=1.2)print(proc.batch_decode(ids, skip_special_tokens=True)[0])
The language/task prompt is baked into the generation config, so no language= argument is needed.
FLEURS + Common Voice Corpus 25.0 (Mozilla, March 2025; https://commonvoice.mozilla.org/en/datasets), capped per the paper. Text-only data from the Goldfish corpus (Chang et al., 2026).
Monolingual (Swahili only). Evaluated on FLEURS / Common Voice test splits; other domains or dialects may differ.
Project page: https://lemn-lab.github.io/buzzasr-docs/