# Bangla Audio Dataset

> A speech and audio collection organized by recording session, with one nested session archive.

## At a Glance
| Metric | Value |
| --- | --- |
| Archives | 1 |
| Total WAV files | 4477 |
| Primary format | Session-organized audio archive |
| Delivery format | Zip archives inside each folder |

## Archive Inventory
| Archive | WAV files | Sessions / notes |
| --- | ---: | --- |
| `Himel Audio.zip` | 4477 | Session-organized audio |

## Folder Layout
```text
Bangla-Audio-Dataset/
+-- Himel Audio.zip
    +-- Session 1/
    |   +-- wavs/
    +-- session 2/
    |   +-- wavs/
    +-- session 3 walid/
    |   +-- Walid.zip
    +-- ...
```

## Working With the Data
1. Extract `Himel Audio.zip`.
2. Use the session folders as the primary organization unit.
3. Review the nested `Walid.zip` package if you need the complete session 3 content.

## Notes
- The archive is organized by recording session rather than by train/validation/test splits.
- File naming is mixed across sessions, so keep the original folder structure intact when copying or extracting.