CML-TTS multilingual speech synthesis dataset
- Provider
- Frederico S. Oliveira et al., Center of Excellence in Artificial Intelligence (CEIA), Federal University of Goias
- License
- CC-BY-4.0 (source)
- Size
- 1,617 files
- Format
- Parquet (in sampled files)
- Stored at
- Hugging Face (https://huggingface.co/datasets/ylacombe/cml-tts)
- Access
- Free and public. No account with the data provider is needed.
- Category
- Speech & Audio
- Source
- https://arxiv.org/abs/2306.10097
About CML-TTS multilingual speech synthesis dataset
CML-TTS (CML-Multi-Lingual-TTS) is a text-to-speech dataset developed at the Center of Excellence in Artificial Intelligence of the Federal University of Goias, Brazil. It comprises audiobooks of public domain Project Gutenberg books read by LibriVox volunteers in seven languages, Dutch, German, French, Italian, Polish, Portuguese and Spanish, all at 24 kHz, and was built from the same LibriVox material as Multilingual LibriSpeech but re-segmented and filtered for speech synthesis. It is used to train and evaluate multilingual and cross-lingual TTS, zero-shot voice cloning and multilingual vocoders in languages with little open TTS data. This repository restructures the OpenSLR SLR146 archives for streaming: one folder per language (dutch, french, german, italian, polish, portuguese, spanish) holding Parquet shards with audio, transcription and speaker metadata, plus the original segments archive (measured 2026-11-14; the Dutch folder alone holds 385 Parquet files; the Hub reports about 836 GB for the repo). Load a language with datasets.load_dataset('ylacombe/cml-tts', 'german', split='train'). Released by its authors on OpenSLR under CC BY 4.0; cite Oliveira et al., 'CML-TTS', 25th International Conference on Text, Speech and Dialogue.
How to load CML-TTS multilingual speech synthesis dataset in Python
from datasets import load_dataset
ds = load_dataset("ylacombe/cml-tts")
print(ds)Data structure
File types in sampled files: Parquet
Files
Example file: dutch/dev-00000-of-00006-3c6a56d39a46e0f1.parquet
Analyze this dataset without downloading it
On Scigantic, CML-TTS multilingual speech synthesis dataset opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.
This page is also available as Markdown and as JSON.
Related Speech & Audio datasets
- Omnilingual Asr Corpus
- VoiceBank+DEMAND noisy speech database (16 kHz)
- Bach Doodle Dataset: 21.6 Million User Melodies with Coconet Four-Voice Bach-Style Harmonizations (Google Magenta)
- CVSS: Multilingual Speech-to-Speech Translation Corpus, 21 Languages into English (CVSS-C and CVSS-T, Google Research)
- Multilingual Librispeech
- E-GMD: Expanded Groove MIDI Dataset, 444 Hours of Human Drumming on 43 Drum Kits with Aligned MIDI and Audio (Google Magenta)
- CocoChorales: 240,000 Synthesized Four-Part Chorale Mixtures with Per-Instrument Stems, MIDI and Note Expression (Google Magenta)
- MInDS-14: multilingual spoken intent detection