|
--- |
|
dataset_info: |
|
features: |
|
- name: audio |
|
dtype: audio |
|
- name: text |
|
dtype: string |
|
- name: length |
|
dtype: int64 |
|
- name: category |
|
dtype: string |
|
splits: |
|
- name: test |
|
num_bytes: 1293604 |
|
num_examples: 24 |
|
download_size: 1293869 |
|
dataset_size: 1293604 |
|
configs: |
|
- config_name: default |
|
data_files: |
|
- split: test |
|
path: data/test-* |
|
license: apache-2.0 |
|
task_categories: |
|
- text-to-speech |
|
language: |
|
- en |
|
size_categories: |
|
- n<1K |
|
--- |
|
|
|
# ElevenLabs Multilingual V2 Technical Speech Dataset |
|
|
|
This dataset contains automatically generated technical phrases in three domains, converted to speech using the ElevenLabs Multilingual V2 model with Adam voice. |
|
|
|
## Dataset Description |
|
|
|
The dataset includes audio samples of technical phrases across three categories: |
|
- Machine Learning (ML) |
|
- Science |
|
- Technology |
|
|
|
Each entry contains: |
|
- Audio file in MP3 format (22050Hz) |
|
- Source text |
|
- Text length |
|
- Category label |
|
|
|
## Data Generation |
|
|
|
The phrases were procedurally generated using predefined templates to create coherent technical statements. All audio was synthesized using: |
|
- Model: ElevenLabs Multilingual V2 |
|
- Voice: Adam |
|
- Voice Settings: |
|
- Stability: 0.5 |
|
- Similarity Boost: 0.75 |
|
- Style: 0.0 |
|
- Speaker Boost: Enabled |
|
|
|
## Sample Entry |
|
|
|
```python |
|
{ |
|
'audio': 'path/to/audio.mp3', |
|
'text': 'The Distributed computing infrastructure leveraging containerized applications efficiently orchestrates high-throughput transactions...', |
|
'length': 195, |
|
'category': 'tech_phrase' |
|
} |
|
``` |
