Software for text rewriting needed, can be without interface

Budget: 4000 UAH Deadline: 4 days

Hello!
I am ready to undertake the implementation of your project.

Recently, I worked on a similar solution — a video processing system with recognition and rephrasing of text. Your task is fully feasible in a local mode:

without third-party services,

using GPU (support for your RTX 4060 will ensure significant acceleration),

support for multiple languages (Russian, English, Spanish, German, Italian, Polish — can be organized, details will be clarified),

result — high-quality text rewriting with an originality level of 90–95% while preserving the meaning and tone of the original.

The program will:
📁 automatically process all videos from the input folder,
📝 extract and recognize speech,
🔁 rephrase and structurally modify the text,
📁 save the result in output.

I would like to clarify:

Average length of videos (you specified up to 12 hours — need to discuss technical limits).

Maximum number of videos per run.

Let's discuss the details privately — I will tell you which options are optimal for local work.

Looking forward to your message! Expand

Budget: 8000 UAH Deadline: 5 days

Hello!
I have carefully reviewed your technical assignment — and I can propose the implementation of an autonomous desktop application for Windows that will perform batch text rewriting from videos, without using third-party services (all locally, with the possibility of acceleration via NVIDIA RTX 4060 GPU).

What will be included in the solution:

Batch processing:

Upload video files to the input folder.

Receive unique texts in the output folder.

Audio extraction and transcription:

Using a local model (for example, Whisper CPP or Faster-Whisper with CUDA support) for high-quality speech recognition, including long videos up to 12 hours.

Supported languages: Russian, English, German, Spanish, Italian, Polish (the list can be expanded upon discussion).

Text rewriting:

Automatic paraphrasing (locally) considering:

Synonymization.

Rearrangement and simplification of sentences.

Preservation of the style and intonation of the original.

Ensuring 90–95% uniqueness.

Using one of the modern language models with local integration (for example, Mistral, Gemma, LLaMA, OpenHermes, or specialized models for paraphrasing) — GPU will be utilized to the maximum.

Simple user interface (optional):

Ability to select input/output language.

Settings for rewriting depth (balance between accuracy and uniqueness).

Progress bar and error log. Expand

Budget: 2000 UAH Deadline: 1 day

Good afternoon!
I am developing software in Python!
I have created similar software for transcribing audio files into text with acceptable processing quality.
I would like to discuss the project details, I am confident I can help you.

Budget: 6000 UAH Deadline: 5 days

The price may be lower or higher depending on your needs.
Write to me to discuss the details.

I have written an approximate stack for this:

Python 3.11+

video/audio processing (ffmpeg, moviepy),

speech recognition (Whisper),

NLP rewriting (LLM, Transformers),

GUI (PyQt, Tkinter),

GPU computations (CUDA via PyTorch).

---------------------

Whisper (offline speech-to-text)
Model: whisper-large-v3 from OpenAI (via CTranslate2 for GPU).

Advantages:

Works locally.

Supports multilingualism (RU, EN, DE, IT, ES, PL, etc.).

GPU acceleration.

Interface: via faster-whisper or ct2-transformers.

---------------------------

Model: Mistral 7B, LLaMA 3, or similar (via GGUF/GGML)
Works offline through llama.cpp, text-generation-webui, or llm CLI.

Format: GGUF (for RAM-efficient generation).

Interface: llama-cpp-python or local server.

Alternative:

Vicuna / Nous Hermes / Zephyr / Mixtral (selected for better rewriting).

Output language — same as input video.

--------------------

ffmpeg (extract audio from video, cut large files)

moviepy (optional — for extracting frames if needed)

-------------------
CUDA + PyTorch — for rewriting and transcription models.

CTranslate2 — allows running Whisper on GPU.

llama.cpp with CUDA/CLBlast — for running LLaMA/Mistral models on GPU.

-------------
Adding automatic language detection from audio (langdetect, whisper autodetect).

Possibility of translation (argos-translate offline if language conversion is needed). Expand

Budget: 1200 UAH Deadline: 3 days

Hello! I have reviewed the project and am ready to start working. I am confident you will be satisfied with the result.