Skip to content

Dataset · n4ze3m

American English synthetic speech corpus

n4ze3m/tiny-voca

19 hours of paired male and female voice recordings for training lightweight TTS models.

What's inside

It contains 17,371 audio clips totaling 19.1 hours in American English, where each sentence is voiced by two synthetic voices: female (F2) and male (M2). The 24 kHz mono recordings (16-bit FLAC) are packaged in parquet files. The data was collected for training the tiny-voca model and is available under the CC BY 4.0 license.

What you can do with it

  • Training or fine-tuning compact text-to-speech (TTS) systems
  • Training voice cloning and voice conversion models
Tasks
Text → speech
Languages
English
Size
10K–100K rows
Downloads this month
874
Likes
66

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("n4ze3m/tiny-voca")