GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
tahaalselwii
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
tahaalselwii
Model Tree
openai/whisper-small
Input Modalities
Output Modalities
Supported Functionality
openai/whisper-smalltahaalselwiiThe pretrained base model and the fine-tuned model were evaluated on the same held-out custom test split.
| Model | Test loss | WER (%) |
|---|---|---|
openai/whisper-small | 3.6458 | 84.3240 |
tahaalselwii/whisper-small-yemeni | 0.9923 | 63.4929 |
The fine-tuned model achieved a 24.70% relative improvement, with WER decreasing from 84.32% to 63.49%, a reduction of 20.83 percentage points.
bash
pip install -U torch transformers accelerate "datasets[audio]"
python
import torchfrom datasets import Audio, load_datasetfrom transformers import AutoModelForSpeechSeq2Seq, AutoProcessormodel_id = "tahaalselwii/whisper-small-yemeni"device = "cuda" if torch.cuda.is_available() else "cpu"processor = AutoProcessor.from_pretrained(model_id)model = AutoModelForSpeechSeq2Seq.from_pretrained(model_id).to(device)model.eval()# Load the Yemeni test splitdataset = load_dataset("UBC-NLP/Casablanca","Yemen",split="test",)# Resample the audio to 16 kHz as required by Whisperdataset = dataset.cast_column("audio",Audio(sampling_rate=16_000),)# Select one test sampleexample = dataset[0]audio = example["audio"]inputs = processor(audio["array"],sampling_rate=audio["sampling_rate"],return_tensors="pt",return_attention_mask=True,)with torch.no_grad():predicted_ids = model.generate(input_features=inputs.input_features.to(device),attention_mask=inputs.attention_mask.to(device),language="ar",task="transcribe",)prediction = processor.batch_decode(predicted_ids,skip_special_tokens=True,)[0]print("Reference:", example["transcription"])print("Prediction:", prediction)
The model was trained using the Yemen configuration of
UBC-NLP/Casablanca. The dataset snapshot used by the training initially contained
1,606 examples across its available source splits. All available source splits were concatenated, shuffled with seed
42, and divided into a new custom split:
| Custom split | Percentage | Examples used |
|---|---|---|
| Train | 80% | 1,284 |
| Validation | 10% | 161 |
| Test | 10% | 160 |
Yemeni Arabic speech data is scarce, and no sufficiently large, suitable dataset was available for this work. As a result, the model was trained and evaluated on a limited dataset, including only 160 test examples. The reported results should therefore be considered preliminary, and performance may vary with new speakers and recording conditions.
If you use this model, please cite the model repository, the Casablanca dataset paper, and the Whisper paper.
bibtex
@misc{tahaalselwii2026whisper_yemeni,author = {Taha Alselwi},title = {Whisper Small Yemeni},year = {2026}}
bibtex
@article{talafha2024casablanca,title = {Casablanca: Data and Models for Multidialectal Arabic Speech Recognition},author = {Talafha, Bashar and others},journal = {arXiv preprint arXiv:2410.04527},year = {2024}}
bibtex
@article{radford2022robust,title = {Robust Speech Recognition via Large-Scale Weak Supervision},author = {Radford, Alec and Kim, Jong Wook and Xu, Tao and Brockman, Greg and McLeavey, Christine and Sutskever, Ilya},journal = {arXiv preprint arXiv:2212.04356},year = {2022}}
This model builds on OpenAI Whisper and the Casablanca dataset created by the UBC Deep Learning and NLP Lab and its contributors.