Update README.md
Browse files
README.md
CHANGED
|
@@ -1,5 +1,75 @@
|
|
| 1 |
-
|
| 2 |
-
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# DataCatalyst Multilingual TTS Sample
|
| 2 |
+
**Hindi | English | Hinglish (Code-Switched)**
|
| 3 |
+
Version: v1.2 | April 2026
|
| 4 |
+
|
| 5 |
+
---
|
| 6 |
+
|
| 7 |
+
## Overview
|
| 8 |
+
|
| 9 |
+
This dataset is a multilingual speech sample curated by DataCatalyst for ASR and TTS evaluation.
|
| 10 |
+
|
| 11 |
+
Languages included:
|
| 12 |
+
- Hindi
|
| 13 |
+
- English (Indian accent)
|
| 14 |
+
- Hinglish (code-switched)
|
| 15 |
+
|
| 16 |
+
Each language contains 6 utterances (~10 seconds each).
|
| 17 |
+
|
| 18 |
+
---
|
| 19 |
+
|
| 20 |
+
## Key Features
|
| 21 |
+
|
| 22 |
+
- 48 kHz, 24-bit mono WAV
|
| 23 |
+
- Speaker-consistent recordings
|
| 24 |
+
- Loudness-normalized variants (EBU R128 aligned)
|
| 25 |
+
- Rich annotations (emotion, style, intent, delivery attributes)
|
| 26 |
+
- Explicit contributor consent
|
| 27 |
+
|
| 28 |
+
---
|
| 29 |
+
|
| 30 |
+
## Dataset Structure
|
| 31 |
+
|
| 32 |
+
/hindi/
|
| 33 |
+
/audio/
|
| 34 |
+
/audio_processed/
|
| 35 |
+
speaker_metadata.json
|
| 36 |
+
utterances.json
|
| 37 |
+
|
| 38 |
+
/english/
|
| 39 |
+
/audio/
|
| 40 |
+
/audio_processed/
|
| 41 |
+
speaker_metadata.json
|
| 42 |
+
utterances.json
|
| 43 |
+
|
| 44 |
+
/hinglish/
|
| 45 |
+
/audio/
|
| 46 |
+
/audio_processed/
|
| 47 |
+
speaker_metadata.json
|
| 48 |
+
utterances.json
|
| 49 |
+
|
| 50 |
+
---
|
| 51 |
+
|
| 52 |
+
## Audio Specifications
|
| 53 |
+
|
| 54 |
+
- Format: WAV (PCM)
|
| 55 |
+
- Sample Rate: 48,000 Hz
|
| 56 |
+
- Bit Depth: 24-bit
|
| 57 |
+
- Channels: Mono
|
| 58 |
+
- Loudness (processed): ~ -22 to -18 LUFS
|
| 59 |
+
- True Peak: ~ -6 to -3 dBTP
|
| 60 |
+
|
| 61 |
+
---
|
| 62 |
+
|
| 63 |
+
## License
|
| 64 |
+
|
| 65 |
+
This dataset is released under the **DataCatalyst Evaluation License (DEL-1.0)**.
|
| 66 |
+
|
| 67 |
+
Permitted:
|
| 68 |
+
- Internal evaluation and benchmarking
|
| 69 |
+
|
| 70 |
+
Not permitted:
|
| 71 |
+
- Commercial use
|
| 72 |
+
- Production model training
|
| 73 |
+
- Redistribution
|
| 74 |
+
|
| 75 |
+
Contact: divyam@datacatalyst.in
|