GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
ifs
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
ifs
Model Tree
openai/whisper-large-v3
Input Modalities
Output Modalities
Supported Functionality
Error rates in percent. CER and WER are computed after lowercasing and removing punctuation; raw CER compares the output exactly as produced.
| Test set | Clips | Speakers | CER | WER | raw CER |
|---|---|---|---|---|---|
| official test split | 4,983 | 258 | 5.5 | 23.6 | 7.0 |
| run 2's test set | 1,122 | 80 | 5.2 | 22.8 | 6.7 |
For comparison, stock Whisper-large-v3 gets 33.0% CER on the official test split. The full comparison is in the write-up.
python
import torchfrom transformers import pipeline asr = pipeline( "automatic-speech-recognition", model="ifs/whisper-large-v3-tt-cv27-run1", dtype=torch.float16, # torch.float32 on a CPU device="cuda", # or "cpu")result = asr( "audio.mp3", return_timestamps=True, # needed for audio longer than 30 seconds generate_kwargs={ "language": "tatar", "task": "transcribe", "num_beams": 1, "no_repeat_ngram_size": 3, "repetition_penalty": 1.2, },)print(result["text"])The decoding settings are the ones used in evaluation; they follow BuzzASR. The weights are stored in float32 (6.2 GB); float16 halves the memory and works well for inference.
All clips come from the validated bucket of Common Voice Scripted Speech 27.0 Tatar. This run kept the official test split untouched, which leaves few speakers for everything else:
| Split | Clips | Speakers | How it was built |
|---|---|---|---|
| test | 4,983 | 258 | the official test split |
| dev | 628 | 2 | two light speakers, one clip per sentence |
| train | 19,605 | 9 | every other validated clip sharing no speaker and no sentence with dev or test |
One speaker accounts for 58% of the training clips. The exact splits are in
the splits/ folder of this repository. Labels keep their
original casing and punctuation.
Full fine-tuning following the BuzzASR recipe, with
finetune.py --max-dev-speaker-clips 350:
Hardware: one NVIDIA A100 80GB on RunPod. Software: PyTorch 2.8.0, Transformers 5.18.0 (full environment in the write-up).
MIT, like the base model. The training data is CC0.
import torchfrom transformers import pipeline asr = pipeline( "automatic-speech-recognition", model="ifs/whisper-large-v3-tt-cv27-run1", dtype=torch.float16, # torch.float32 on a CPU device="cuda", # or "cpu")result = asr( "audio.mp3", return_timestamps=True, # needed for audio longer than 30 seconds generate_kwargs={ "language": "tatar", "task": "transcribe", "num_beams": 1, "no_repeat_ngram_size": 3, "repetition_penalty": 1.2, },)print(result["text"])