ananddey/torongo-tts-as
Text-to-Speech • 36.3M • Updated • 40
Datasets:
Assamese Single Female Speech Dataset
A high-quality open curated single-speaker Assamese speech dataset for Text-to-Speech (TTS), voice synthesis, speech generation and language research.This dataset contains paired Assamese neutral speech recordings and text transcripts.
Splits
| Split | Samples |
|---|---|
| train | 11318 |
| val | 231 |
Total audio: 20.53 hours.
Fields
| Field | Type | Description |
|---|---|---|
id |
string | Stable per-clip identifier |
audio |
Audio | Waveform |
text |
string | Raw Assamese transcript |
speaker_id |
string | Speaker identifier |
gender |
string | Speaker gender |
language |
string | as (Assamese) |
duration |
float | Clip duration in seconds |
sampling_rate |
int | Native sampling rate of the stored audio |
License
CC-BY-4.0.
Maintainer
Anand Dey
Contributions, feedback, and collaborations are welcome.
- Downloads last month
- 14
Models trained or fine-tuned on ananddey/assamese-single-fem-dataset