transloco-ita-lld

Italian to Ladin Real-Time Translation Model

This is a fast, lightweight real-time translation model from Italian (it) to Ladin (lld), based on Helsinki-NLP/opus-mt-itc-itc and optimized using CTranslate2 for efficient inference.

💡 Key Features

• ✅ Base model: Helsinki-NLP/opus-mt-itc-itc
• ⚡ Optimized with CTranslate2
• 🧠 int8 quantization for faster inference and lower memory usage
• 🗣️ Designed for real-time transcription + translation use cases (e.g., TransLoco)
• 🕒 Suitable for low-latency environments like live subtitling or in-browser translation tools

🏗️ Model Architecture

• Architecture: Transformer
• Format: CTranslate2
• Quantization: int8
• Size on disk: ~70 MB

🚀 Intended Use

• Real-time speech-to-speech or speech-to-text translation from Italian to Ladin
• Assistive tools for minority language accessibility
• Educational and research applications
• Use as part of tools like TransLoco

Non-commercial use only, in accordance with the CC BY-NC 4.0 license.

import ctranslate2
from transformers import AutoTokenizer

mtmodel = ctranslate2.Translator("./transloco-ita-lld", device="cpu")
tokenizer = AutoTokenizer.from_pretrained("./transloco-ita-lld")

texts = ["Questo è un esempio."]
tokenized_sentences = [tokenizer.convert_ids_to_tokens(tokenizer.encode(x)) for x in texts]

batch_res = mtmodel.translate_batch(source=tokenized_sentences)

decoded_results = [
     tokenizer.decode(
         tokenizer.convert_tokens_to_ids(res.hypotheses[0]),
         skip_special_tokens=True
     ) for res in batch_res
]

print(decoded_results)

⚠️ Note: The tokenizer uses fur_Latn as the target language code due to the lack of lld_Latn support in the original NLLB vocabulary.

❗Limitations

• Ladin is a low-resource language, and the model may struggle with:
  • Out-of-domain vocabulary
  • Variant-specific variations
• The model may hallucinate outputs when given incomplete or noisy input.

⚖️ Ethical Considerations

• Language technologies for minority languages should be developed with community involvement.
• Please avoid using the model for commercial applications or mass-translation pipelines without review.

📎 Citation

If you use this model in your work, please cite:

@incollection{Haller-Seeber_Frontull_2026,
  author    = {Haller-Seeber, Simon and Frontull, Samuel},
  title     = {TransLoco: AI-Driven Real-Time Transcription, Translation, and Summarization: A self-hosted free-software conference solution},
  booktitle = {Explorations at the Intersections of Media, Education, and Philosophy: (2nd extended edition)},
  editor    = {Rummler, Klaus and Pallaver, G{\"u}nther and Missomelius, Petra and Dander, Valentin and Leistert, Oliver},
  publisher = {OAPublishing Collective},
  address   = {Z{\"u}rich, Switzerland},
  year      = {2026},
  month     = {feb},
  pages     = {397--418},
  series    = {Medien -- Wissen -- Bildung},
  doi       = {10.21240/978-3-03978-164-5_21},
  url       = {https://www.oapublishing.ch/omp/index.php/oap/catalog/book/132/chapter/137},
  abstract  = {Advanced AI models, such as Large Language Models (LLMs), require significant computational resources and are thus typically hosted on centralized cloud infrastructure. This implies data protection risks and dependencies on major platforms, such as those of tech monopolies. This paper presents TransLoco, a self-hosted, free-software framework that enables near real-time transcription and translation of audio streams locally on one’s own devices. We discuss practical free-software tools and models, along with their hardware requirement specifications, and demonstrate how advanced AI applications can be deployed without any dependency on external resources. This allows for greater security, control, and full digital sovereignty. Moreover, as part of our effort to support linguistic diversity, we include the low-resource language Ladin and release an Italian-Ladin translation model optimized for fast inference. The framework was demonstrated and tested live at the Medien-Wissen-Bildung 2025 conference, confirming its setup and practical applicability in real-world scenarios. We made the complete source code publicly available at https://git.uibk.ac.at/informatik/iis/iis-projects/TransLoco.}
}