Update README.md
Browse files
README.md
CHANGED
|
@@ -62,3 +62,63 @@ configs:
|
|
| 62 |
- split: train
|
| 63 |
path: images/train-*
|
| 64 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 62 |
- split: train
|
| 63 |
path: images/train-*
|
| 64 |
---
|
| 65 |
+
# Vaani Sample Data — Andhra Pradesh / Annamaya
|
| 66 |
+
|
| 67 |
+
A small filtered slice of the [ARTPARK-IISc/VAANI](https://huggingface.co/datasets/ARTPARK-IISc/VAANI) dataset, covering the `AndhraPradesh_Annamaya` subset. Only rows that have **both a transcript and a speaker ID** are retained. The reference images cited by those rows are bundled as a second config.
|
| 68 |
+
|
| 69 |
+
## Configs
|
| 70 |
+
|
| 71 |
+
| Config | Split | Rows | Description |
|
| 72 |
+
|----------|-------|-----:|-------------|
|
| 73 |
+
| `audio` | train | 1,302 | Telugu speech utterances with transcripts and speaker metadata. |
|
| 74 |
+
| `images` | train | 973 | Reference images cited via `referenceImage` in the audio config (973 unique images across the 1,302 audio rows). |
|
| 75 |
+
|
| 76 |
+
## Loading
|
| 77 |
+
|
| 78 |
+
```python
|
| 79 |
+
from datasets import load_dataset
|
| 80 |
+
|
| 81 |
+
audio = load_dataset("SujithPulikodan/Vaani-sample-data", "audio", split="train")
|
| 82 |
+
images = load_dataset("SujithPulikodan/Vaani-sample-data", "images", split="train")
|
| 83 |
+
```
|
| 84 |
+
|
| 85 |
+
## Schema
|
| 86 |
+
|
| 87 |
+
### `audio`
|
| 88 |
+
|
| 89 |
+
| Column | Type | Notes |
|
| 90 |
+
|---|---|---|
|
| 91 |
+
| `audio` | `Audio` | Embedded audio (WAV bytes + path). |
|
| 92 |
+
| `file_name` | `string` | Audio basename. |
|
| 93 |
+
| `transcript` | `string` | Telugu transcription. May contain `<noise>` markers. |
|
| 94 |
+
| `speakerID` | `string` | Stable speaker identifier. |
|
| 95 |
+
| `language` | `string` | Spoken language (Telugu). |
|
| 96 |
+
| `gender` | `string` | Speaker gender. |
|
| 97 |
+
| `state`, `district`, `pincode` | `string` | Recording location. |
|
| 98 |
+
| `duration` | `float32` | Seconds. |
|
| 99 |
+
| `languagesKnown` | `string` | Stringified list, e.g. `['Telugu']`. |
|
| 100 |
+
| `stay_years` | `string` | How long the speaker has lived in the district. |
|
| 101 |
+
| `isTranscriptionAvailable` | `string` | `Yes`/`No` (always `Yes` after filtering). |
|
| 102 |
+
| `referenceImage` | `string` | Path of the prompt image shown to the speaker (matches a row in the `images` config). |
|
| 103 |
+
| `speakerImageHash` | `string` | Hash of the speaker's image. |
|
| 104 |
+
| `UtteranceSequenceID` | `int32` | Sequence index within a session. |
|
| 105 |
+
|
| 106 |
+
### `images`
|
| 107 |
+
|
| 108 |
+
| Column | Type | Notes |
|
| 109 |
+
|---|---|---|
|
| 110 |
+
| `image` | `Image` | JPEG bytes. |
|
| 111 |
+
| `file_name` | `string` | Image basename, matches the basename in `audio.referenceImage`. |
|
| 112 |
+
|
| 113 |
+
To join an audio row with its image:
|
| 114 |
+
|
| 115 |
+
```python
|
| 116 |
+
import os
|
| 117 |
+
imgs_by_name = {row["file_name"]: row["image"] for row in images}
|
| 118 |
+
def image_for(audio_row):
|
| 119 |
+
return imgs_by_name[os.path.basename(audio_row["referenceImage"])]
|
| 120 |
+
```
|
| 121 |
+
|
| 122 |
+
## Source & License
|
| 123 |
+
|
| 124 |
+
Derived from **ARTPARK-IISc/VAANI** by IISc / ARTPARK. Refer to the upstream dataset card for the original license and terms of use; this redistribution inherits them. Please cite the upstream project when using this data.
|