Propicto/propicto-orfeo

Propicto-orféo

📝 Dataset Description

Propicto-orféo is a dataset of aligned speech-id/transcription/pictograms (the pictograms correspond to the identifier associated with an ARASAAC pictogram) in French. It was created from the CEFC-Orféo corpus. This dataset was presented in the research paper titled "A Multimodal French Corpus of Aligned Speech, Text, and Pictogram Sequences for Speech-to-Pictogram Machine Translation" at LREC-Coling 2024. The dataset was split into training, validation, and test sets.

Propicto-orféo contains three CSV files : train, valid and test, with the following statistics :

Split	Number of utterances
train	231 374
valid	28 796
test	29 009

• Curated by: Cécile MACAIRE
• Funded by : PROPICTO ANR-20-CE93-0005
• Language(s) (NLP): French
• License: CC-BY-NC-SA-4.0

⚒️ Dataset Structure

Each file contains the following information :

id : the unique identifier of the utterance, which corresponds to a unique audio clip file (in wav) for the orféo dataset
text : the transcription of the audio clip
pictos : the sequence of id pictograms from ARASAAC
tokens : the sequence of tokens, each of them is the keyword associated to the ARASAAC id pictogram

💡 Dataset example

For the given sample :

id : cefc-cfpb-1000-5-1186
text : tu essayes de mélanger les deux
pictos : [6625, 26144, 7074, 5515, 5367]
tokens : toi essayer de mélanger à_côté_de

• The clip is from the Orféo subcorpus CFPB, 1000-5, with the sentence ID 1186.
  
• The text is the associated transcription, in en : “you try to mix the two”.
  
• Pictos is the sequence of pictogram IDs, each of them can be retrieved from here : 6625 = https://static.arasaac.org/pictograms/6625/6625_2500.png
  
• Tokens are retrieved from a specific lexicon and can be used to train translation models.

Dataset Sources

• Repository: CEFC-Orféo
• Papers :
  • C. Benzitoun, J.-M. Debaisieux, H.-J. Deulofeu (2016). Le projet ORFÉO : un corpus d'études pour le français contemporain. Corpus n°15, p. 91-114
  • J.-M. Debaisieux & C. Benzitoun (2020). Orféo : un corpus et une plateforme pour l’étude du français contemporain. Langages n°219

💻 Uses

Propicto-orféo is intended to be used to train Speech-to-Pictograms translation and Text-to-Pictograms translation models. This dataset can also be used to fine-tune large language models to perform translation into pictograms.

⚙️ Dataset Creation

The dataset is created by applying a specific formalism that converts french oral transcriptions into a corresponding sequence of pictograms.
The formalism includes a set of grammatical rules to handle specific phenomenon (negation, name entities, pronominal form, plural, ...) to the French language, as well as a dictionary which associates each ARASAAC ID pictogram with a set of keywords (tokens).
This formalism was presented at LREC.

Source Data : conversations / meetings / daily life situations (oral transcriptions)

Bias, Risks, and Limitations

The translation can be partially incorrect, due to incorrect or missing words translated into pictograms.

📌 Citation

@inproceedings{macaire-etal-2024-multimodal,
    title = "A Multimodal {F}rench Corpus of Aligned Speech, Text, and Pictogram Sequences for Speech-to-Pictogram Machine Translation",
    author = "Macaire, C{\'e}cile  and
      Dion, Chlo{\'e}  and
      Arrigo, Jordan  and
      Lemaire, Claire  and
      Esperan{\c{c}}a-Rodier, Emmanuelle  and
      Lecouteux, Benjamin  and
      Schwab, Didier",
    booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)",
    year = "2024",
    publisher = "ELRA and ICCL",
    url = "https://aclanthology.org/2024.lrec-main.76",
    pages = "839--849",
}

@inproceedings{macaire24_interspeech,
  title     = {Towards Speech-to-Pictograms Translation},
  author    = {Cécile Macaire and Chloé Dion and Didier Schwab and Benjamin Lecouteux and Emmanuelle Esperança-Rodier},
  year      = {2024},
  booktitle = {Interspeech 2024},
  pages     = {857--861},
  doi       = {10.21437/Interspeech.2024-490},
  issn      = {2958-1796},
}

👩‍🏫 Dataset Card Authors

Cécile MACAIRE, Chloé DION, Emmanuelle ESPÉRANÇA-RODIER, Benjamin LECOUTEUX, Didier SCHWAB