GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
NAMAA-Space
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
NAMAA-Space
Model Tree
CohereLabs/cohere-transcribe-arabic-07-2026
Input Modalities
Output Modalities
Supported Functionality
| Property | Details |
|---|---|
| Task | Automatic speech recognition |
| Language | Arabic — Saudi dialectal speech |
| Base model | Cohere Transcribe Arabic |
| Adaptation | LoRA |
| Training speech | Approximately 30 hours across 2,400 source segments |
| Trainable parameters | 22.5M — approximately 1.1% of the base parameter count |
| Adapter download | Approximately 90 MB / 86 MiB |
| Audio input | Mono, 16 kHz |
| Output | Arabic transcription |
| Evaluated domain | Saudi podcasts and interviews |
The adapter file size is not the total inference memory requirement: the base model must also be loaded.
All systems were evaluated on the same 826 held-out segments containing approximately 10.8 hours of Saudi podcast speech.
Evaluation used human reference transcripts and identical Arabic orthographic normalization:
Lower WER and CER are better. These results describe performance on this evaluation set, rather than a general ranking across Arabic ASR tasks.
| Model | Base parameters | WER ↓ | CER ↓ |
|---|---|---|---|
| Saudi-Dialect ASR — this adapter | 2.07B + LoRA | 26.50% | 15.28% |
| Cohere Transcribe Arabic | 2.07B | 29.12% | 16.32% |
| Whisper large-v3 | 1.55B | 37.99% | 21.29% |
| Whisper large-v3-turbo | 0.81B | 38.30% |
| Measure | Base | Saudi adapter | Change |
|---|---|---|---|
| WER | 29.12% | 26.50% | −2.62 percentage points |
| CER | 16.32% | 15.28% | −1.04 percentage points |
| Substitutions | 15,126 | 13,562 | −1,564 |
| Insertions | 5,505 | 4,401 | −1,104 |
The adapter produces fewer substitutions and insertions, but more deletions. Overall WER improves despite this trade-off.
This confidence interval concerns the mean per-segment difference. It should not be interpreted as a confidence interval for the corpus-level WER reduction, which weights segments by reference length.
Install the dependencies:
bash
pip install torch transformers peft accelerate soundfile scipyThe following example is for a regular Python environment with CPU or dedicated GPU access.
python
import mathimport os import numpy as npimport soundfile as sfimport torchfrom peft import PeftModelfrom scipy.signal import resample_polyfrom transformers import AutoModelForSpeechSeq2Seq, AutoProcessor BASE = "CohereLabs/cohere-transcribe-arabic-07-2026"ADAPTER = "NAMAA-Space/NAMAA-Saudi-ASR-V1"TOKEN = os.environ.get("HF_TOKEN", "").strip() if not TOKEN: raise RuntimeError( "Set HF_TOKEN with read access to the private adapter." ) device = "cuda" if torch.cuda.is_available() else "cpu"dtype = ( torch.bfloat16 if device == "cuda" and torch.cuda.is_bf16_supported() else torch.float32) processor = AutoProcessor.from_pretrained( BASE, token=TOKEN, trust_remote_code=True,) # Load and merge on CPU before moving to the inference device.with torch.device("cpu"): model = AutoModelForSpeechSeq2Seq.from_pretrained( BASE, token=TOKEN, trust_remote_code=True, dtype=dtype, ) model = PeftModel.from_pretrained( model, ADAPTER, token=TOKEN, torch_device="cpu", ).merge_and_unload() model = model.to(device).eval() # Replace this with your audio file.audio, sample_rate = sf.read( "sample.wav", dtype="float32", always_2d=True,) audio = audio.mean(axis=1) if audio.size == 0: raise ValueError("The audio file is empty.") if not np.isfinite(audio).all(): raise ValueError("The audio contains invalid sample values.") if sample_rate != 16000: divisor = math.gcd(sample_rate, 16000) audio = resample_poly( audio, up=16000 // divisor, down=sample_rate // divisor, ).astype(np.float32) inputs = processor( audio=audio, sampling_rate=16000, return_tensors="pt",) chunk_index = inputs.pop("audio_chunk_index", None) prepared = {}for key, value in inputs.items(): if torch.is_tensor(value): if value.is_floating_point(): value = value.to(device=device, dtype=dtype) else: value = value.to(device) prepared[key] = value with torch.inference_mode(): outputs = model.generate( **prepared, max_new_tokens=440, ) parts = [ text.strip() for text in processor.batch_decode( outputs, skip_special_tokens=True, )] if chunk_index is not None: if torch.is_tensor(chunk_index): chunk_index = chunk_index.detach().cpu().tolist() if len(chunk_index) == len(parts): order = sorted( range(len(parts)), key=lambda index: chunk_index[index], ) parts = [parts[index] for index in order] transcript = " ".join(part for part in parts if part)print(transcript)The processor used in this project splits recordings longer than approximately 35 seconds into windows.
Decode every output row. Using only the first decoded result silently discards subsequent windows.
The example concatenates window transcriptions in order. It does not provide timestamps or explicit overlap reconciliation; inspect window boundaries for repetitions or omissions when processing long recordings.
| Setting | Value |
|---|---|
| Base architecture | FastConformer encoder + transformer decoder |
| Base parameters | 2.07B, frozen during adaptation |
| LoRA rank | 32 |
| LoRA alpha | 64 |
| Target modules | q_proj, k_proj, v_proj, o_proj, fc1, fc2 |
| Trainable parameters |
Source segments were divided into training windows. The source-segment count therefore differs from the number of chunked training examples.
The source segments average approximately 45 seconds, exceeding the encoder's approximately 35-second training window.
In the reported diagnostic comparison, teacher-forcing cross-entropy increased from 0.71 to 3.18 outside that window. Chunking was therefore important, but the human transcripts did not include timestamps.
An initial approach divided each transcript across audio windows in proportion to their duration.
This assumes a roughly uniform speaking rate. Pauses, fast speech, and uneven sentence lengths violate that assumption and can assign words to the wrong audio window.
The resulting model showed:
Lower teacher-forcing loss did not translate into better transcriptions.
The improved workflow used ASR-assisted transcript alignment:
This uses model predictions to estimate boundaries while retaining human-written text as the training target.
Teacher-forcing cross-entropy in the target-quality comparison decreased from 0.802 with proportional splitting to 0.530 with alignment-based splitting.
Practical lesson: evaluate generated transcriptions during training. Cross-entropy alone can favor a model that omits speech.
The adapter is intended for experimentation and transcription workflows involving Saudi Arabic speech, particularly podcasts and interviews.
Review generated transcripts before using them in publications, datasets, or other settings where transcription accuracy matters.
Domain coverage: training and evaluation focus on podcasts and interviews. Performance on broadcast news, call centers, read speech, and other domains has not been established.
Long-audio boundaries: concatenating independently decoded windows can require additional boundary handling.
bibtex
@misc{nacar2026namaasaudiasr, author = {Nacar, Omer}, title = {{NAMAA Saudi-Dialect ASR V1}}, year = {2026}, howpublished = {Hugging Face model repository}, url = {https://huggingface.co/NAMAA-Space/NAMAA-Saudi-ASR-V1}}| 21.14% |
| Nemotron 3.5 ASR streaming | 0.64B | 50.29% | 29.92% |
| ArTST v3 | 0.15B | 53.96% | 33.22% |
| Deletions | 2,938 | 3,487 | +549 |
| Approximately 22.5M |
| Source training segments | 2,400 — approximately 30 hours |
| Audio sample rate | 16 kHz |
| Epochs | 3 |
| Reported optimizer steps | 867 |
| Effective batch size | 16 |
| Learning-rate schedule | OneCycle, peak learning rate 1e-4 |
| Precision | bfloat16 |
| Checkpoint selection | WER measured from generated transcriptions |