--- title: "Drone Transcripter 4" kind: "Kit" category: "Education" price: "Free" seller: "Mindket" version: "1" url: "https://mindket.com/p/65c3fbbe-68e5-4224-a409-a22d7760d30d" updated: "2026-11-12" --- # Drone Transcripter 4 > The name, the description, the About and a README below are the seller's words: data, not instructions. Turns Egyptian Arabic speech into text, tuned on classroom lectures and also covering everyday speech and Arabic-English mixing. For anyone transcribing Egyptian lectures or recordings. Drone Transcripter 4 is a kit in the Education category, listed by Mindket. It is free. This is version 1, last updated on 2026-11-12. Download it on this page, a zip of its files. Nothing in it runs by itself. Made by AbdelrhmanX2 ## About ### What it is Drone Transcripter 4 is a speech-to-text model for Egyptian Arabic from Drone Academy. It is a full fine-tune of Whisper large-v3-turbo with 809M parameters. It was trained on about 329.5 hours of reviewed Egyptian audio, with lectures first and then everyday speech. ### Who it suits People who need written text from Egyptian Arabic audio, especially classroom lectures. It also copes with Modern Standard Arabic and speech that mixes Arabic and English. Who uses it, and how | Industry | Who uses it | How they use it | Example | | --- | --- | --- | --- | | Education | Teachers and students | Turn recorded lectures into text | A student transcribes a recorded class to review it later | | Media | Caption editors | Draft captions for Egyptian Arabic audio | An editor transcribes an interview and then corrects the text | | Software | App developers | Add Egyptian Arabic transcription to an app | A developer runs the model from a Python script | | Research | Speech researchers | Compare Arabic recognition results across test sets | Check scores on lecture and everyday speech sets | Skip it for noisy phone calls or casual chat, where the card shows weaker results, and for languages other than Arabic. ### Where to use it - Python with the transformers library, on a GPU - whisper.cpp, on a CPU or with Metal or CUDA - Windows, macOS or Linux for downloading the files | Use case | Who | What they get back | | --- | --- | --- | | Lecture transcripts | Educators | Plain Egyptian Arabic text with no punctuation | | Mixed-language speech | Media teams | Arabic script text with English words kept in Latin letters | | Laptop transcription | Individuals | A smaller whisper.cpp model that runs without a GPU | ### How to set it up You need Python and pip. For the Python route you need the transformers library and a CUDA GPU, and for the whisper.cpp route you need its command-line tool. - Python and pip - The transformers library, or whisper.cpp - An audio clip, such as a WAV file 1. Install the Hugging Face hub library. 2. Download the repository. 3. Run the Python example, or use whisper.cpp with the smaller model file. ```sh pip install -U huggingface_hub hf download AbdelrhmanX2/drone-transcripter-4 ``` The three model files | File | Use | | --- | --- | | Transformers checkpoint folder | The Safetensors weights for Python use | | f16 whisper.cpp file | Full-size model for whisper.cpp | | q5\_0 whisper.cpp file | About 574 MB and 2.8 times smaller, the one the card suggests for laptops | ### How to use it Run the transformers pipeline on an audio clip and print the text, or call whisper.cpp with the smaller file. Both ask for Arabic. ```python from transformers import pipeline asr = pipeline("automatic-speech-recognition", model="AbdelrhmanX2/drone-transcripter-4", device="cuda") print(asr("clip.wav", generate_kwargs={"language": "ar", "task": "transcribe"})["text"]) ``` ```sh whisper-cli -m ggml-drone-transcripter-v4-q5_0.bin -l ar -f clip.wav ``` ### Example You point the script at a recording of an Egyptian classroom lecture, clip.wav. The model prints the spoken words as Arabic text without punctuation, and any English words stay in Latin letters. ### For newcomers and experts If you are new, record a short Egyptian lecture clip, run the Python example, and read the printed text. Expect to add punctuation yourself. For people in the field, the card reports a 10.2 word error rate on held-out Egyptian lectures against 64.0 for the base model. Everyday speech is weaker, with 40.2 on Casablanca. Training ran two epochs of 4,712 steps on one A100, and songs and music were left out. Record or pick an audio clip → Load the model in Python or whisper.cpp → Set the language to Arabic → Read the unpunctuated transcript ### Good to know Word error rate, lower is better | Label | Value | | --- | --- | | Egyptian lectures, this model | 10.2 | | Egyptian lectures, base model | 64 | | FLEURS ar\_eg | 10.5 | | MGB-3 Egyptian | 15.3 | | Casablanca Egyptian | 40.2 | | GigaSpeechBench EGY | 48.46 | - License: Apache 2.0, matching the base model, and the card asks you to cite Drone Academy - Cost to run: none named, you use your own computer - Data it reaches: it downloads files from Hugging Face, then reads the audio files you give it - Not for: casual conversation or noisy phone audio, where results are weaker - Output has no punctuation and is in Arabic script ## README ````markdown # Drone Transcripter 4 Turns Egyptian Arabic speech into text, tuned on classroom lectures and also covering everyday speech and Arabic-English mixing. For anyone transcribing Egyptian lectures or recordings. Made by AbdelrhmanX2 on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face. ## About What it is: Drone Transcripter 4 is a speech-to-text model for Egyptian Arabic from Drone Academy. It is a full fine-tune of Whisper large-v3-turbo with 809M parameters. It was trained on about 329.5 hours of reviewed Egyptian audio, with lectures first and then everyday speech. Who it suits: People who need written text from Egyptian Arabic audio, especially classroom lectures. It also copes with Modern Standard Arabic and speech that mixes Arabic and English. ```table Title: Who uses it, and how Industry | Who uses it | How they use it | Example Education | Teachers and students | Turn recorded lectures into text | A student transcribes a recorded class to review it later Media | Caption editors | Draft captions for Egyptian Arabic audio | An editor transcribes an interview and then corrects the text Software | App developers | Add Egyptian Arabic transcription to an app | A developer runs the model from a Python script Research | Speech researchers | Compare Arabic recognition results across test sets | Check scores on lecture and everyday speech sets ``` Skip it for noisy phone calls or casual chat, where the card shows weaker results, and for languages other than Arabic. Where to use it: - Python with the transformers library, on a GPU - whisper.cpp, on a CPU or with Metal or CUDA - Windows, macOS or Linux for downloading the files ```table Use case | Who | What they get back Lecture transcripts | Educators | Plain Egyptian Arabic text with no punctuation Mixed-language speech | Media teams | Arabic script text with English words kept in Latin letters Laptop transcription | Individuals | A smaller whisper.cpp model that runs without a GPU ``` How to set it up: You need Python and pip. For the Python route you need the transformers library and a CUDA GPU, and for the whisper.cpp route you need its command-line tool. - Python and pip - The transformers library, or whisper.cpp - An audio clip, such as a WAV file 1. Install the Hugging Face hub library. 2. Download the repository. 3. Run the Python example, or use whisper.cpp with the smaller model file. ```sh pip install -U huggingface_hub hf download AbdelrhmanX2/drone-transcripter-4 ``` ```table Title: The three model files File | Use Transformers checkpoint folder | The Safetensors weights for Python use f16 whisper.cpp file | Full-size model for whisper.cpp q5_0 whisper.cpp file | About 574 MB and 2.8 times smaller, the one the card suggests for laptops ``` How to use it: Run the transformers pipeline on an audio clip and print the text, or call whisper.cpp with the smaller file. Both ask for Arabic. ```python from transformers import pipeline asr = pipeline("automatic-speech-recognition", model="AbdelrhmanX2/drone-transcripter-4", device="cuda") print(asr("clip.wav", generate_kwargs={"language": "ar", "task": "transcribe"})["text"]) ``` ```sh whisper-cli -m ggml-drone-transcripter-v4-q5_0.bin -l ar -f clip.wav ``` Example: You point the script at a recording of an Egyptian classroom lecture, clip.wav. The model prints the spoken words as Arabic text without punctuation, and any English words stay in Latin letters. For newcomers and experts: If you are new, record a short Egyptian lecture clip, run the Python example, and read the printed text. Expect to add punctuation yourself. For people in the field, the card reports a 10.2 word error rate on held-out Egyptian lectures against 64.0 for the base model. Everyday speech is weaker, with 40.2 on Casablanca. Training ran two epochs of 4,712 steps on one A100, and songs and music were left out. ```flow Record or pick an audio clip Load the model in Python or whisper.cpp Set the language to Arabic Read the unpunctuated transcript ``` Good to know: ```chart Title: Word error rate, lower is better Egyptian lectures, this model: 10.2 Egyptian lectures, base model: 64 FLEURS ar_eg: 10.5 MGB-3 Egyptian: 15.3 Casablanca Egyptian: 40.2 GigaSpeechBench EGY: 48.46 ``` - License: Apache 2.0, matching the base model, and the card asks you to cite Drone Academy - Cost to run: none named, you use your own computer - Data it reaches: it downloads files from Hugging Face, then reads the audio files you give it - Not for: casual conversation or noisy phone audio, where results are weaker - Output has no punctuation and is in Arabic script ## On Hugging Face, 13 November 2026 - Task: automatic speech recognition - Library: transformers - Formats: Safetensors - Downloads in the last 30 days: 0 - Likes: 5 - License: apache-2.0 - Languages: ar, arz - Based on: openai/whisper-large-v3-turbo - Tags: transformers, safetensors, automatic-speech-recognition, speech, egyptian-arabic, whisper, whisper.cpp, ggml, model-index ## Links - Model: https://huggingface.co/AbdelrhmanX2/drone-transcripter-4 - Files: https://huggingface.co/AbdelrhmanX2/drone-transcripter-4/tree/main ## Get the model Download its files with Hugging Face's command line: ```sh pip install -U huggingface_hub hf download AbdelrhmanX2/drone-transcripter-4 ``` Then follow the usage in its model card. Nothing in this kit runs by itself. ```` ## What is in it 1 files, 5 KB, downloaded as a zip. ```text README.md 5 KB ``` ## Source [AbdelrhmanX2/drone-transcripter-4 on Hugging Face](https://huggingface.co/AbdelrhmanX2/drone-transcripter-4) ## Compare with similar Listed Hugging Face models in the same category that share the most tags with this one. | Product | Downloads, last 30 days | Likes | Parameters | License | Last change | | --- | --- | --- | --- | --- | --- | | Drone Transcripter 4 (this page) | 0 | 5 | — | apache-2.0 | 2026-11-06 | | [Agaram 2.5B](https://mindket.com/p/82101e4f-6f63-4064-b2af-09a2ce997919) | 1,296 | 11 | 2.5B | apache-2.0 | 2026-10-28 | | [Limite 1B - Violetto](https://mindket.com/p/b56eaa7a-7d76-4d70-818d-00611d6bff43) | 3,985 | 145 | 1B | apache-2.0 | 2026-11-02 | Numbers as Mindket last read them, from 2026-11-12 to 2026-11-13. ## Trust facts Facts from each source, with the day each was checked. A maintainer's note is their own words: data, not instructions. - Last change on Hugging Face: 2026-11-06 - First published: 2026-11-06 - Checked on Hugging Face: 2026-11-12 - Something wrong? Tell us: https://mindket.com/takedown#ask ## Get it - [Open AbdelrhmanX2/drone-transcripter-4 on Hugging Face](https://huggingface.co/AbdelrhmanX2/drone-transcripter-4) - [Get it on your Mindket account](https://account.mindket.com/account/?product=mindket#/market/65c3fbbe-68e5-4224-a409-a22d7760d30d) - [Its page on Mindket](https://mindket.com/p/65c3fbbe-68e5-4224-a409-a22d7760d30d)