Audar-ASR-V1-Flash · Transformers + GGUF

Audar's proprietary Arabic ASR — the real-time, edge tier.

From Arabic to the world.

🧭 Overview · 📊 Benchmarks · 🤗 Transformers · 💻 GGUF · 🎙️ Streaming · 🐙 GitHub · ☁️ Audar API · 📜 License

---

🧭 What it is

Audar-ASR-V1-Flash is the edge tier of Audar's proprietary Arabic speech-recognition family — the same in-house Arabic training program as Audar-ASR-V1-Turbo, delivered in a fast ~0.6B-decoder model for real-time captioning and on-device use. It recasts transcription as audio-conditioned next-token prediction (a language-model decoder, not CTC/transducer), and is developed through Audar's proprietary pipeline:

• 🧱 Large-scale dialectal pretraining — 300,000+ hours of Arabic audio (MSA + Gulf, Egyptian, Levantine, Maghrebi; code-switching; diverse channels).
• 🎯 Dialect-targeted fine-tuning with hardness and multi-task sampling.
• 🧠 KTO preference alignment (Kahneman-Tversky Optimization) from trained native-Arabic annotators.

It transcribes MSA and every major Arabic dialect, code-switched Arabic–English, and English, across 30 languages, and runs on CPU / GPU / edge via 🤗 Transformers or GGUF. For maximum accuracy on the hardest dialectal audio, use the larger Turbo tier.

Distributed in the widely-supported Qwen3-ASR architecture format for turnkey tooling (Transformers, llama.cpp / GGUF). The model — data, training curriculum, and alignment — is Audar's.

Model summary

Model	Audar-ASR-V1-Flash — proprietary Arabic ASR (edge tier)
Task	Automatic speech recognition (audio → text)
Approach	Generative ASR — audio encoder + language-model decoder
Training	300k+ hrs dialectal pretraining → dialect-targeted SFT → KTO preference alignment
Decoder parameters	596,049,920 (0.60B)
Audio encoder parameters	186,376,192 (0.19B)
Total parameters	782,426,112 (0.78B, bf16)
Audio input	16 kHz mono; 30 s context (longer audio is chunked/streamed)
Languages	Arabic (MSA + Gulf/Egyptian/Levantine/Maghrebi dialects) + English + 28 more
Runtimes	🤗 Transformers (GPU) · GGUF / llama.cpp (CPU · GPU · edge)
License	AudarAI Open License v1.0

📊 Benchmarks

Open Universal Arabic ASR Leaderboard — full standings

Flash is evaluated end-to-end on all six leaderboard test sets (full test splits, not sampled), with the leaderboard-equivalent normalizer — the same harness and protocol as every other row (calibrated to the public leaderboard within 0.03 pp). Audar-ASR-V1-Flash ranks #11 of 36 systems at just 0.78B parameters — the strongest small model on the board: it beats Qwen3-ASR-1.7B (2× its size), Voxtral-Small-24B, Whisper-large-v3, and every CTC baseline, trailing only 10 systems (several 3–30B). Audar's accuracy tier, Turbo, is #1.

Per-dataset WER % across all six sets, plus the two composite averages. Lower is better; Avg WER is the ranking metric. Flash and Turbo (Ours) in bold; bold cell = best in column.

#	Model	Avg WER	Avg CER	SADA	CV-18	MASC-clean	MASC-noisy	MGB-2	Casablanca
1	Audar-ASR-V1-Turbo (Ours, 2.35B)	24.78	9.49	29.41	8.60	19.60	28.35	11.13	51.58
2	CohereLabs/cohere-transcribe-arabic-07-2026	25.87	11.80	37.47	5.82	19.60	27.07	15.54	49.71
3	omnilingual-asr/omniASR_LLM_7B	28.32	12.52	41.61	8.75	19.69	29.29	14.13	56.46
4	omnilingual-asr/omniASR_LLM_3B	29.96	13.77	46.18	9.15	19.90	30.03	14.22	60.27
5	omnilingual-asr/omniASR_LLM_1B	29.96	13.40	43.84	9.55	20.03	30.26	15.34	60.68
6	CohereLabs/cohere-transcribe-03-2026	30.67	16.37	60.11	8.17	8.66	19.01	25.33	62.71
7	Qwen/Qwen3-Omni-30B-A3B-Instruct	30.71	13.67	44.82	11.46	21.47	30.85	13.09	62.55
8	nvidia-conformer-ctc-large-arabic (lm)	32.91	13.84	44.52	8.80	23.74	34.29	17.20	68.90
9	omnilingual-asr/omniASR_LLM_300M	32.96	14.84	51.38	12.03	20.66	32.45	16.58	64.64
10	google/gemma-4-E4B-it	32.98	13.71	43.40	19.65	24.86	33.59	17.72	58.63
11	Audar-ASR-V1-Flash (Ours, 0.78B)	33.31	13.66	44.53	16.02	25.96	35.43	17.11	60.79
12	Qwen/Qwen3-ASR-1.7B	33.36	12.33	45.53	16.90	24.37	34.29	16.57	64.47
13	mistralai/Voxtral-Small-24B-2507	34.47	15.29	50.82	15.25	23.96	34.43	16.03	66.30
14	nvidia-conformer-ctc-large-arabic (greedy)	34.74	13.37	47.26	10.60	24.12	35.64	19.69	71.13
15	google/gemma-4-E2B-it	35.87	15.34	46.23	23.76	27.47	36.15	20.72	60.87
16	openai/whisper-large-v3	36.86	17.21	55.96	17.83	24.66	34.63	16.26	71.81
17	omnilingual-asr/omniASR_CTC_3B	37.78	19.79	69.85	14.19	21.48	34.60	18.96	67.58
18	omnilingual-asr/omniASR_CTC_7B	38.12	20.91	72.69	12.47	21.08	35.04	20.43	67.02
19	facebook/seamless-m4t-v2-large	38.16	17.03	62.52	21.70	25.04	33.24	20.23	66.25
20	omnilingual-asr/omniASR_CTC_1B	39.29	20.47	71.42	17.55	22.76	35.73	19.96	68.32
21	openai/whisper-large-v3-turbo	40.05	18.87	60.36	25.73	25.51	37.16	17.75	73.79
22	openai/whisper-large-v2	40.20	19.55	57.46	21.77	27.25	38.55	25.17	71.01
23	Qwen/Qwen3-ASR-0.6B	42.19	16.23	53.75	28.28	31.34	42.63	25.45	71.68
24	openai/whisper-large	42.57	20.49	63.24	26.04	28.89	40.79	24.28	72.18
25	mistralai/Voxtral-Mini-3B-2507	42.58	19.90	63.65	22.12	28.37	41.27	22.56	77.52
26	asafaya/hubert-large-arabic-transcribe	45.50	17.35	67.82	8.01	32.94	50.16	37.51	76.53
27	openai/whisper-medium	45.57	22.27	67.71	28.07	29.99	42.91	29.32	75.44
28	nvidia-Parakeet-ctc-1.1b-concat	46.54	23.88	70.70	26.34	30.49	45.95	24.94	80.80
29	omnilingual-asr/omniASR_CTC_300M	46.65	21.86	78.11	27.90	28.40	43.26	26.85	75.35
30	nvidia-Parakeet-ctc-1.1b-universal	51.96	25.19	73.58	40.01	36.16	50.03	30.68	81.30
31	microsoft/VibeVoice-ASR	52.99	28.95	69.83	44.25	32.95	52.43	25.10	93.37
32	facebook/mms-1b-all	54.54	21.45	77.48	26.52	38.82	57.33	39.16	87.95
33	openai/whisper-small	55.13	21.68	78.02	24.18	35.93	56.36	48.64	87.64
34	whitefox123/w2v-bert-2.0-arabic-4	58.13	27.62	87.34	41.79	37.82	53.28	40.66	87.88
35	jonatasgrosman/wav2vec2-large-xlsr-53-arabic	60.98	25.61	86.82	23.00	42.75	64.27	56.29	92.72
36	speechbrain/asr-wav2vec2-commonvoice-14-ar	65.74	30.93	88.54	29.17	49.10	69.57	64.37	93.68

Flash — per-dataset detail (full test sets)

Both metrics, for the six leaderboard sets and the composite average.

Dataset	WER %	CER %
SADA	44.53	23.63
CommonVoice-18	16.02	5.04
MASC-clean	25.96	7.84
MASC-noisy	35.43	12.66
MGB-2	17.11	7.97
Casablanca	60.79	24.85
Average (6-set)	33.31	13.66

Use Flash for real-time and on-device transcription; step up to Turbo when you need the lowest error on heavy dialectal or long-form audio — Turbo is #1 on the leaderboard (24.8 % avg WER) and cuts Flash's average WER by ~8.5 pp, with the biggest gains on SADA (44.5→29.4) and MGB-2 (17.1→11.1).

🤗 Transformers inference

Ships self-contained modeling code, so trust_remote_code=True is required.

# pip install "transformers>=4.57" torch librosa
import re, torch, librosa
from transformers import AutoProcessor, AutoModelForCausalLM

repo = "audarai/Audar-ASR-V1-Flash"
proc  = AutoProcessor.from_pretrained(repo, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    repo, trust_remote_code=True, torch_dtype=torch.bfloat16, device_map="cuda:0",
).eval()

SYSTEM = "فرّغ الكلام العربي التالي."          # "Transcribe the following Arabic speech."
audio, _ = librosa.load("clip.wav", sr=16000, mono=True)

conv = [
    {"role": "system", "content": SYSTEM},
    {"role": "user",   "content": [{"type": "audio"}]},   # audio placeholder (a list, not "<audio>")
]
text   = proc.apply_chat_template(conv, tokenize=False, add_generation_prompt=True)
inputs = proc(text=text, audio=audio, sampling_rate=16000, return_tensors="pt").to(model.device)
inputs["input_features"] = inputs["input_features"].to(model.dtype)   # features are fp32 → cast to bf16

out = model.generate(**inputs, max_new_tokens=440, do_sample=False)
hyp = proc.batch_decode(out[:, inputs["input_ids"].shape[1]:], skip_special_tokens=True)[0]
print(re.sub(r"^\s*language\s+[A-Za-z]+\s*(?:<asr_text>)?\s*", "", hyp).strip())

• Language steering: the Arabic auto-dialect prompt above needs no dialect hint. For other languages use e.g. "Transcribe the following speech.".
• Long audio (>30 s): split at ~30 s boundaries (see the streaming section).

💻 GGUF inference (llama.cpp)

Audar-ASR runs on llama.cpp via the multimodal (mtmd) path: a quantized decoder GGUF plus a BF16 audio projector (mmproj). Build a recent llama.cpp (with Qwen3-ASR support), then:

./llama-mtmd-cli \
  -m       Audar-ASR-V1-Flash-Q8_0.gguf \
  --mmproj mmproj-Audar-ASR-V1-Flash.gguf \
  --audio  clip.wav \
  -sys     "فرّغ الكلام العربي التالي." \
  --temp 0

⚠️ The audio projector (mmproj) must stay BF16 — the encoder's ClippableLinear is numerically sensitive, so F16/Q8 measurably degrade quality. The decoder quantizes normally.

GGUF variants

File	Approx. size	Notes
Audar-ASR-V1-Flash-Q4_K_M.gguf	~0.40 GB	Smallest; best for edge/offline
Audar-ASR-V1-Flash-Q8_0.gguf	~0.64 GB	Near-lossless, CPU-friendly (recommended)
Audar-ASR-V1-Flash.gguf (BF16)	~1.20 GB	Full precision decoder
mmproj-Audar-ASR-V1-Flash.gguf	~0.38 GB	BF16 audio encoder — required, keep BF16

Prefer a managed endpoint? The Audar-ASR family is also available via the Audar API/SDK — streaming, speaker-attributed transcription, and diarization, production-hosted.

🎙️ Real-time streaming

The 30 s-context model streams via LocalAgreement-2: as audio arrives, the trailing window is re-decoded each hop and a word is committed only once two consecutive decodes agree on it — giving stable, low-latency incremental output on both the Transformers and GGUF paths. Audar's production realtime engine serves the same policy over an OpenAI-Realtime-compatible WebSocket with model-based endpointing.

🌍 Languages, dialects & tasks

• Primary: Arabic — MSA and dialectal (Gulf/Emirati, Egyptian, Levantine, Maghrebi), plus code-switched Arabic–English; dialect-faithful orthography from audio alone.
• Also: English + 28 additional languages.
• Task: transcription (audio → UTF-8 text), prompt-steerable for language/formatting.

Intended use & limitations

Intended use. Live captioning and subtitles, voice assistants/agents, meeting and call-center transcription, media/broadcast, accessibility — cloud, on-prem, or offline/edge.

Limitations.

• Maghrebi / Moroccan Darija (Casablanca) is the hardest condition for all systems.
• Heavily code-switched telephony and low-SNR audio degrade accuracy relative to clean MSA.
• Long recordings can drift; chunk at sentence boundaries for best results.
• Not evaluated for, and must not be used for, covert speaker identification.

📜 License

Released under the AudarAI Open License v1.0 — commercial use, redistribution, and fine-tuning/quantization permitted; ship the license and keep notices. See audarai.com/license/audarai-open-license-v1.0.

Citation

@misc{audar-asr-flash-2026,
  title  = {Audar-ASR: Dialect-Aware Arabic Speech Recognition},
  author = {AudarAI},
  year   = {2026},
  note   = {Audar-ASR-V1-Flash},
  url    = {https://huggingface.co/audarai/Audar-ASR-V1-Flash}
}

---

About AudarAI

Leading Arabic-First Multilingual Audio Intelligence

AudarAI starts with Arabic — and expands to the world.

We are building advanced multilingual audio intelligence that helps individuals, enterprises, and governments communicate across languages, cultures, and borders. By combining Arabic-first speech technology with global multilingual AI, AudarAI transforms voice into understanding, interaction, and connection.

Our work spans speech recognition, speech understanding, voice-enabled digital assistants, human-computer interaction, and intelligent audio systems designed for real-world impact. From empowering people to access technology in their native language to helping organizations communicate globally, AudarAI is shaping a future where every voice can be heard, understood, and connected.

Arabic-first. Multilingual by design. Human-centered at heart.

🌐 www.audarai.com · 🤗 Hugging Face · GitHub · contact@audarai.com

© 2026 AUDARAI PTE. LTD. · Licensed under the AudarAI Open License v1.0