---
language:
- en
task_categories:
- text-to-speech
tags:
- audio
- tts
- maya1
- expressive-speech
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*.parquet
---

# test-v3

Synthetic expressive English speech generated with Maya1 (Q8_0). 34 clips, 0.11 hours, each 8-15 seconds, 24000 Hz mono 16-bit WAV.

Columns: `id`, `audio` (WAV), `text` (with inline tags such as `<laugh>`), `voice_description` (Maya1 voice prompt), `duration` (seconds).

```python
from datasets import load_dataset
ds = load_dataset("Shadow0482/test-v3", split="train")
```