International Multilingual Speech Dataset (international-multilingual)

A large-scale 7-language multilingual speech dataset containing audio recordings and CSV transcriptions for German (DE), English (UK & US), Spanish (ES), French (FR), Italian (IT), Polish (PL), and Russian (RU). Features gender-divided metadata for balanced model training.

Dataset Description

This dataset provides speech audio and aligned text transcriptions across 7 languages, with both standard locale-level CSV files and gender-specific breakdowns in the csvs/ subdirectory. The audio data comes as compressed WAV archives per language/locale. Ideal for training multilingual ASR systems, cross-lingual speech models, and gender-aware acoustic models.

Languages Included

Language	Locale/Code	Male CSV	Female CSV	Audio Archive	Text CSV
German	de_DE	✅	✅	de_DE_wavs.zip	de_DE.csv
English (UK)	en_UK	—	—	en_UK_wavs.zip	en_UK.csv
English (US)	en_US	✅	✅	en_US_wavs.zip	en_US.csv
Spanish	es_ES	✅	✅	es_ES_wavs.zip	es_ES.csv
French	fr_FR	✅	✅	fr_FR_wavs.zip	fr_FR.csv
Italian	it_IT	✅	✅	it_IT_wavs.zip	it_IT.csv
Polish	pl_PL	✅	✅	pl_PL_wavs.zip	pl_PL.csv
Russian	ru_RU	✅	✅	ru_RU_wavs.zip	ru_RU.csv

Use Cases

• Multilingual ASR: Train a single model to transcribe speech across 7 languages
• Gender-balanced Training: Use the per-gender CSV files to control for speaker gender distribution
• Cross-lingual Transfer Learning: Pre-train on high-resource languages, fine-tune on low-resource ones
• Accented Speech Recognition: Distinguish between UK and US English varieties
• Voice Cloning & TTS: Source data for multilingual text-to-speech synthesis

File Inventory

Directory/File	Description	Size
de_DE_wavs.zip	German (Germany) audio recordings	—
en_UK_wavs.zip	English (UK) audio recordings	—
en_US_wavs.zip	English (US) audio recordings	—
es_ES_wavs.zip	Spanish (Spain) audio recordings	—
fr_FR_wavs.zip	French (France) audio recordings	—
it_IT_wavs.zip	Italian (Italy) audio recordings	—
pl_PL_wavs.zip	Polish (Poland) audio recordings	—
ru_RU_wavs.zip	Russian (Russia) audio recordings	—
de_DE.csv	German CSV with transcriptions	—
en_UK.csv	English (UK) CSV with transcriptions	—
en_US.csv	English (US) CSV with transcriptions	—
es_ES.csv	Spanish CSV with transcriptions	—
fr_FR.csv	French CSV with transcriptions	—
it_IT.csv	Italian CSV with transcriptions	—
pl_PL.csv	Polish CSV with transcriptions	—
ru_RU.csv	Russian CSV with transcriptions	—
csvs/	Directory with gender-split CSVs (male/female per locale)	—
.gitattributes	Git LFS configuration	—
README.md	Dataset documentation (this file)	—

Usage

Loading the Dataset

from datasets import load_dataset

dataset = load_dataset("edwixx/international-multilingual", trust_remote_code=True)

Extracting by Language

# Example: extract German audio
unzip de_DE_wavs.zip -d de_DE_audio/

# View German transcriptions
cat de_DE.csv

Accessing Gender-based Splits

Gender-divided metadata is available under the csvs/ directory:

# View male vs female CSVs
ls csvs/

# Example: German female speakers
head csvs/de_DE_female.csv

Access Control

This dataset is gated and requires manual approval. Request access at: https://huggingface.co/datasets/edwixx/international-multilingual

Metadata

• Author: Anurag Kanade
• Languages: German, English (UK/US), Spanish, French, Italian, Polish, Russian
• License: MIT
• Hosted by: Hugging Face Datasets
• Created: 2024-11-21
• Last Updated: 2026-06-25
• Storage Size: ~85.9 GB
• Downloads: 1

Citation

@misc{edwixx-international-multilingual,
  author = {Anurag Kanade},
  title = {international-multilingual},
  year = {2026},
  publisher = {Hugging Face},
  journal = {Hugging Face Hub},
  howpublished = {\url{https://huggingface.co/datasets/edwixx/international-multilingual}}
}