BroadRouter

Arabic AI articles

Updated

Arabic speech to text and Arabic LLMs: dialects, accuracy, cost per audio minute, and how to test them on your own audio and text. 26 articles, each opening with a direct answer. Every outside fact links to its source with the date we checked it.

  1. 8 min read · Updated

    Add diacritics to Arabic text with an LLM: prompt, checks, DER

    Ask the model for full tashkeel and nothing else, then strip the diacritics from its reply and compare with your input. Any changed letter means reject and retry. Score a hand-checked sample with diacritic error rate (DER) and word error rate (WER), with and without case endings. Published results put GPT-4 close to, but behind, the best dedicated diacritizers.

  2. 8 min read · Updated

    Arabic grammar and spelling correction with an LLM: a how-to

    Ask the model for the corrected text only, with the fewest possible edits and no rephrasing. Then diff the reply against the input in code: reject replies that changed too much, and label each word-level edit as hamza, taa marbuta, alef maqsura, merge, split or other. Score a hand-corrected sample on edits, with precision, recall and F0.5.

  3. 6 min read · Updated

    Arabic meeting minutes from recordings: an API pipeline

    To make Arabic meeting minutes with an API, transcribe the recording with an Arabic speech-to-text model, split long transcripts into sections, ask an LLM for decisions, action items and open questions as JSON, then merge the sections. Have the model quote the transcript for every action item, and write the minutes in Modern Standard Arabic or English.

  4. 7 min read · Updated

    Arabic named entity extraction with an LLM: JSON, dates and checks

    Ask the model to copy each entity exactly as written, with a type from a fixed list, as JSON. Check every entity in code: it must appear in the source once diacritics are stripped and alef forms unified. Convert Hijri dates with a library such as hijridate, not the model. Score a hand-labelled sample with entity-level precision, recall and F1.

  5. 8 min read · Updated

    Arabic OCR API comparison: Google, Azure, Textract, Mistral, LLMs

    Four OCR APIs list Arabic printed text in their own docs: Google Cloud Vision, Azure AI Document Intelligence, Amazon Textract and Mistral OCR. Google, Azure Read and Textract each charge USD 1.50 per 1,000 pages at entry volume. Only Azure lists Arabic handwriting. Vision LLMs, such as Qwen3.8 27B on BroadNet, bill a page as tokens and return the fields you ask for.

  6. 7 min read · Updated

    Arabic subtitles (SRT) from a speech-to-text API: what you need

    An SRT file needs text plus a start and end time for every caption. APIs with word or segment timestamps, or direct SRT export, give the best timing. BroadNet Transcribe Arabic returns text only, so split audio at pauses, transcribe each chunk and time captions from the chunk offsets. Fine for talks and podcasts, not for frame-accurate film subtitles.

  7. 7 min read · Updated

    Arabic text classification with an LLM: sentiment, intent, triage

    To classify Arabic text with an LLM, define a short label set with one-line definitions, give a few labelled examples in each dialect you see, ask for JSON with one label and a short quote as evidence, and set reasoning low. Then score the model on 200 hand-labelled messages with macro F1 before trusting it. Fine-tuned Arabic models can still beat a general LLM.

  8. 7 min read · Updated

    Arabic text normalization for search and RAG: what to change, what not

    For search and RAG indexes, normalize both the stored copy and every query the same way: Unicode NFKC, strip tashkeel and tatweel, unify alef forms, map ى to ي and ة to ه, and convert digits. Keep the original text for display and for the LLM. Treat hamza seats and ta marbuta as tunable: they raise recall but merge some distinct words.

  9. 7 min read · Updated

    Arabic text summarization with an LLM: prompts, length and scoring

    Name the register (usually Modern Standard Arabic, even for dialect input) and a word range in the prompt. Then check in code: count the words, confirm every number in the summary appears in the source, and retry once with the problem stated. Score with an Arabic-aware ROUGE tokenizer plus human review; default rouge-score gives Arabic 0.

  10. 7 min read · Updated

    Design an AI experience for multiple languages

    Store two settings: the interface locale and the content language, and never infer one from the other. Set direction from the locale, isolate user and model text with dir="auto" or bdi, format numbers and dates with the locale, normalise digits on input, treat language detection as a hint, and evaluate quality and cost per language.

  11. 7 min read · Updated

    Translate app strings to Arabic with an LLM: placeholders and plurals

    Send the strings as JSON with every placeholder swapped for a marker, ask for all six Arabic plural forms (zero, one, two, few, many, other), and build the ICU plural in code, not in the model. Validate in CI: same keys, same arguments and tags, all six forms present. Wrap inserted names in Unicode isolates so right-to-left text displays correctly.

  12. 8 min read · Updated

    Translate Arabic audio to English text with an API: a tested pipeline

    Run two steps. Transcribe the Arabic audio with a speech-to-text API, setting language to ar, then translate the transcript with an LLM that is told to keep names and numbers. Split long files first, check numbers in code, and score a small sample with chrF to see how much the transcription step costs you in quality.

  13. 5 min read · Updated

    Arabic ASR word error rate (WER) explained

    Word error rate (WER) is the number of substituted, deleted and inserted words divided by the words in the reference transcript. Lower is better. Arabic WER is hard to compare because diacritics, alef and ta marbuta spelling and dialects inflate it. BroadNet reports 3.8 WER on FLEURS-ar and 23.4 on dialect speech for BroadNet Transcribe Arabic.

  14. 5 min read · Updated

    Arabic dialect speech recognition: Gulf, Egyptian, Levantine, Maghrebi

    Arabic dialects differ from Modern Standard Arabic in sounds, words and spelling, and speakers mix in English or French, so ASR models trained mostly on formal Arabic score far worse on them. To test a provider, build a test set per dialect from your real audio, fix a spelling convention, and score WER, CER and entity accuracy per dialect.

  15. 5 min read · Updated

    Arabic OCR vs vision LLMs: reading Arabic documents and invoices

    Classic OCR turns Arabic page images into raw text; a vision LLM reads the image and returns the fields you ask for, such as vendor, date and total, as JSON. For invoices and receipts a vision LLM usually saves the parsing step. On BroadNet Inference, Qwen3.8 27B, DeepSeek V4.1 Flash and BroadNet Gemma 4 VL 26B accept images.

  16. 5 min read · Updated

    Arabic prompt engineering: register, RTL, numerals and mixed text

    Good Arabic output comes from explicit instructions: name the register (MSA or a specific dialect), the audience, the numeral style (Western 0–9 or Eastern Arabic ٠–٩), the formatting allowed, and how to treat English terms. Render replies with dir="auto" or isolation marks so mixed Arabic and English displays in the right order.

  17. 6 min read · Updated

    Arabic RAG: normalisation, chunking and retrieval that works

    Arabic RAG works when you normalise text before indexing (unify alef forms, ya and alef maqsura, ta marbuta, strip diacritics and tatweel), chunk by paragraph, combine a multilingual embedding model with keyword search, and let a long-context LLM answer from the retrieved passages with citations. Apply the same normalisation to queries.

  18. 7 min read · Updated

    Arabic speech-to-text API: how to choose one in 2026

    Pick an Arabic speech-to-text API by testing it on your own recordings: measure word error rate on your dialects, then compare price per audio hour, file limits and data retention. BroadNet Transcribe Arabic costs $0.25 per audio hour ($0.0042 a minute), and the Open Universal Arabic ASR Leaderboard listed it #4 of 39 on 15 November 2026.

  19. 5 min read · Updated

    Arabic WhatsApp chatbot architecture: voice notes to replies

    An Arabic WhatsApp assistant has five parts: a webhook that receives messages from the WhatsApp Business Platform, a queue, speech to text for voice notes, an LLM that drafts the reply, and a sender that answers through the API. Converting a voice note and transcribing it with BroadNet Transcribe Arabic costs $0.0042 per audio minute.

  20. 5 min read · Updated

    Arabic–English translation with LLMs: prompts, glossaries, evaluation

    To translate Arabic and English with an LLM, give the model the audience, register and a glossary of fixed terms, ask it to keep numbers and names unchanged, and translate in paragraph-sized chunks. Evaluate with chrF or COMET plus native review. In BroadNet's tests, Qwen3.8 27B translated accurately between Arabic and English.

  21. 5 min read · Updated

    Arabizi and LLMs: handling Franco-Arabic with numbers like 3 and 7

    Arabizi, also called Franco-Arabic, is Arabic written in Latin letters with digits for sounds Latin lacks: 3 for ع, 7 for ح, 2 for ء, 5 for خ. LLMs usually understand it but convert it inconsistently. Detect it, ask the model to transliterate into Arabic script with the dialect named, then process the Arabic text.

  22. 7 min read · Updated

    Choosing an LLM API for Arabic: what to test

    Choose an LLM for Arabic by testing it on your own tasks: understanding Gulf, Egyptian and Levantine dialects, writing fluent Modern Standard Arabic, translating Arabic and English, and reading Arabic documents. In BroadNet's tests, Qwen3.8 27B wrote fluent Arabic and understood Gulf and Egyptian dialects. It costs $0.175 per 1M input tokens.

  23. 6 min read · Updated

    Gulf, Egyptian and Levantine Arabic: how LLMs handle dialects

    LLMs handle Arabic dialects unevenly: they usually understand Egyptian and Gulf Arabic better than they write them, and quality drops for rarer dialects. Test understanding and writing separately on your own messages. In BroadNet's own tests, Qwen3.8 27B understood Gulf and Egyptian dialects and wrote fluent Arabic. It costs $0.175 per 1M input tokens.

  24. 4 min read · Updated

    How to transcribe call center recordings: an Arabic-ready pipeline

    To transcribe call center recordings, split stereo files into agent and customer channels, cut each channel into speech turns, transcribe every turn with the right language code, then merge the turns by start time. Redact card numbers and phone numbers from the transcript before an LLM summarises the call. Mono recordings need a separate diarization step.

  25. 6 min read · Updated

    Multilingual speech-to-text API: 14 languages, one endpoint

    BroadNet Transcribe is a speech-to-text model for 14 languages, including English, Spanish, French, German, Chinese, Japanese and Arabic, served on an OpenAI-compatible /v1/audio/transcriptions endpoint. It costs $0.002 per audio minute ($0.12 per hour). For mainly Arabic audio, BroadNet Transcribe Arabic is the specialised option at $0.0042 a minute.

  26. 4 min read · Updated

    Voice notes to text for business: WhatsApp and Telegram to tickets

    To turn customer voice notes into tickets, download each note from the WhatsApp Cloud API or Telegram Bot API as soon as the webhook arrives, convert it to WAV, transcribe it with the right language code, then ask an LLM to extract intent, urgency and order numbers as JSON. Create the ticket with the transcript attached and reply in text.

Read as MarkdownAsk ChatGPTAsk ClaudeAsk Perplexity