GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
zainab11
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
zainab11
Model Tree
openai/whisper-small
Input Modalities
Output Modalities
Supported Functionality
| Model | Raw WER | Normalized WER | Normalized CER | RTF (T4, fp16) |
|---|---|---|---|---|
| openai/whisper-small (zero-shot) | 33.07% | 28.02% | 8.51% | 0.079 |
| this model | 25.82% | 19.90% | 6.77% | 0.077 |
| openai/whisper-large-v3-turbo (zero-shot, reference) | 16.61% | 10.10% | 3.13% | 0.050 |
Normalization: NFC, diacritics/tatweel removal, alef & ya unification, punctuation stripping — applied identically to references and predictions. One undecodable test sample excluded (documented).
Evaluated zero-shot on a 400-clip subset of Casablanca Egypt test (seed 42) — real spontaneous
Egyptian dialect, never seen in training. All clips ≤26s (duration-audited: no truncation).
| Model | Norm WER | Norm CER | Insertion rate |
|---|---|---|---|
| whisper-small (zero-shot) | 103.3% | 76.1% | 0.330 |
| this model | 81.2% | 50.7% | 0.066 |
| whisper-large-v3-turbo (zero-shot) | 71.8% | 35.9% | 0.197 |
Three takeaways: (1) no off-the-shelf model — including turbo — reaches usable accuracy on spontaneous Egyptian; (2) fine-tuning on read MSA still transferred a 21-point WER gain, driven almost entirely by decoding discipline (insertions collapsed 0.33 → 0.07 — the lowest hallucination rate of the three, below even turbo); (3) dialect WER partly reflects non-standardized dialect orthography — CER is the kinder acoustic view. Full analysis: phase 4 report.
ar_eg train (2,102 clips ≤30s ≈ 6h); labels = dataset's transcription fieldAcoustic substitutions improved most (16/19 samples), numbers 7/7, named entities 9/12; rare technical terms unchanged. Number-format gains partly reflect adaptation to FLEURS's orthographic conventions and may not transfer out of domain.
python
# pip install transformers torch librosa soundfileimport torchimport librosafrom transformers import WhisperProcessor, WhisperForConditionalGeneration# 1. Load the modelMODEL_ID = "zainab11/whisper-small-ar-fleurs"processor = WhisperProcessor.from_pretrained(MODEL_ID)model = WhisperForConditionalGeneration.from_pretrained(MODEL_ID)device = "cuda" if torch.cuda.is_available() else "cpu"model = model.to(device).eval()# 2. Load any audio file (WAV/MP3/...) — converted to mono 16 kHz automaticallyaudio_path = "my_audio.wav" # <-- put your file hereaudio, sr = librosa.load(audio_path, sr=16000, mono=True)print(f"Audio duration: {len(audio)/sr:.2f}s")# 3. Audio -> log-Mel features (+ attention mask)inputs = processor(audio, sampling_rate=16000,return_attention_mask=True, return_tensors="pt")input_features = inputs.input_features.to(device)attention_mask = inputs.attention_mask.to(device)# 4. Transcribe (Arabic)with torch.no_grad():ids = model.generate(input_features,attention_mask=attention_mask,language="ar", task="transcribe")# 5. Decodeprint(processor.batch_decode(ids, skip_special_tokens=True)[0])
Note: clips longer than 30s are truncated by this simple example (Whisper's window). For long audio, use the chunked pipeline or segment first.
Fine-tuned on read MSA speech with Egyptian-accented speakers — not spontaneous Egyptian dialect, noisy telephony, or code-switched speech. Inherits Whisper's known long-form repetition/hallucination behavior. As measured on dialectal data above, the model is not suitable for spontaneous Egyptian without further dialect fine-tuning
Part of an end-to-end Arabic ASR study (baselines → normalization → error analysis → fine-tuning): github.com/zainabayman11/whisper-ar-skeleton
🎙️ Try it live: whisper-ar-demo