ITALIC: An Italian Intent Classification Dataset ITALIC is a dataset of Italian audio recordings and contains annotation for utterance transcripts and associated intents. The ITALIC dataset was created through a custom web platform, utilizing both native and non-native Italian speakers as participants. The participants were required to record themselves while reading a randomly sampled short text from the MASSIVE dataset. ITALIC dataset containing 16,521 audio recordings collected by 70 different volunteers. The dataset is composed of: recordings: a folder containing the audio recordings in .wav format. It contains all the recordings composing the data collection. [CONFIG_NAME]_[SPLIT_NAME].json: the files containing metadata used for generating the configuration proposed in the paper and their corresponding splits: [CONFIG_NAME] is the name of the configuration, e.g. massive, hard_noisy, or hard_speaker. For the description of the configurations, please refer to the paper. [SPLIT_NAME] is the name of the split, e.g. train, validation, or test. Each split is different for each configuration. The metadata files are in JSON format, with one sample per line. Each sample is a JSON object with the following fields: id: the unique identifier of the sample. age: the age of the speaker (self-reported) gender: the gender of the speaker (self-reported) region: the region of origin of the speaker (self-reported) nationality: the nationality of the speaker (self-reported) lisp: the presence of a lisp in the speaker (self-reported) education: the education level of the speaker (self-reported) speaker_id: the unique identifier of the speaker environment: the environment in which the recording was made (self-reported) device: the device used for recording (self-reported) scenario, field, intent: the information parsed from massive annotations and accompanying metadata. utt: the utterance to be spoken by the speaker. This information is also taken from massive. Important Note: By downloading and accessing the dataset, you agree not to attempt to determine the identity of speakers in the ITALIC dataset or to clone their voices. License The ITALIC dataset is released under the Creative Commons Attribution 4.0 International License. If you use the dataset in your work, please cite the ITALIC paper.

ITALIC: An Italian Intent Classification Dataset ITALIC is a dataset of Italian audio recordings and contains annotation for utterance transcripts and associated intents. The ITALIC dataset was created through a custom web platform, utilizing both native and non-native Italian speakers as participants. The participants were required to record themselves while reading a randomly sampled short text from the MASSIVE dataset. ITALIC dataset containing 16,521 audio recordings collected by 70 different volunteers. The dataset is composed of: recordings: a folder containing the audio recordings in .wav format. It contains all the recordings composing the data collection. [CONFIG_NAME]_[SPLIT_NAME].json: the files containing metadata used for generating the configuration proposed in the paper and their corresponding splits: [CONFIG_NAME] is the name of the configuration, e.g. massive, hard_noisy, or hard_speaker. For the description of the configurations, please refer to the paper. [SPLIT_NAME] is the name of the split, e.g. train, validation, or test. Each split is different for each configuration. The metadata files are in JSON format, with one sample per line. Each sample is a JSON object with the following fields: id: the unique identifier of the sample. age: the age of the speaker (self-reported) gender: the gender of the speaker (self-reported) region: the region of origin of the speaker (self-reported) nationality: the nationality of the speaker (self-reported) lisp: the presence of a lisp in the speaker (self-reported) education: the education level of the speaker (self-reported) speaker_id: the unique identifier of the speaker environment: the environment in which the recording was made (self-reported) device: the device used for recording (self-reported) scenario, field, intent: the information parsed from massive annotations and accompanying metadata. utt: the utterance to be spoken by the speaker. This information is also taken from massive. Important Note: By downloading and accessing the dataset, you agree not to attempt to determine the identity of speakers in the ITALIC dataset or to clone their voices. License The ITALIC dataset is released under the Creative Commons Attribution 4.0 International License. If you use the dataset in your work, please cite the ITALIC paper.