---
dataset_name: am-pranav/stt-unified-bench
pretty_name: STT Unified Mini-Bench
task_categories:
- automatic-speech-recognition
task_ids:
- speech-recognition
license: other
size_categories:
- n<10K
configs:
- config_name: assamese
  data_files:
  - split: val
    path: assamese/val-*
- config_name: bodo
  data_files:
  - split: val
    path: bodo/val-*
- config_name: dogri
  data_files:
  - split: val
    path: dogri/val-*
- config_name: en_indian_accent
  data_files:
  - split: val
    path: en_indian_accent/val-*
- config_name: gujarati
  data_files:
  - split: val
    path: gujarati/val-*
- config_name: hindi
  data_files:
  - split: val
    path: hindi/val-*
- config_name: kannada
  data_files:
  - split: val
    path: kannada/val-*
- config_name: kashmiri
  data_files:
  - split: val
    path: kashmiri/val-*
dataset_info:
- config_name: assamese
  features:
  - name: audio
    dtype:
      audio:
        sampling_rate: 16000
        decode: false
  - name: text
    dtype: string
  - name: source
    dtype: string
  - name: id
    dtype: string
  splits:
  - name: val
    num_bytes: 186681046.0
    num_examples: 1000
  download_size: 182177487
  dataset_size: 186681046.0
- config_name: bodo
  features:
  - name: audio
    dtype:
      audio:
        sampling_rate: 16000
        decode: false
  - name: text
    dtype: string
  - name: source
    dtype: string
  - name: id
    dtype: string
  splits:
  - name: val
    num_bytes: 169987794.0
    num_examples: 1000
  download_size: 166335246
  dataset_size: 169987794.0
- config_name: dogri
  features:
  - name: audio
    dtype:
      audio:
        sampling_rate: 16000
        decode: false
  - name: text
    dtype: string
  - name: source
    dtype: string
  - name: id
    dtype: string
  splits:
  - name: val
    num_bytes: 250328972.0
    num_examples: 1000
  download_size: 243221303
  dataset_size: 250328972.0
- config_name: en_indian_accent
  features:
  - name: audio
    dtype:
      audio:
        sampling_rate: 16000
        decode: false
  - name: text
    dtype: string
  - name: source
    dtype: string
  - name: id
    dtype: string
  splits:
  - name: val
    num_bytes: 178921558.0
    num_examples: 1000
  download_size: 163516587
  dataset_size: 178921558.0
- config_name: gujarati
  features:
  - name: audio
    dtype:
      audio:
        sampling_rate: 16000
        decode: false
  - name: text
    dtype: string
  - name: source
    dtype: string
  - name: id
    dtype: string
  splits:
  - name: val
    num_bytes: 236196428.0
    num_examples: 1000
  download_size: 231117224
  dataset_size: 236196428.0
- config_name: hindi
  features:
  - name: audio
    dtype:
      audio:
        sampling_rate: 16000
        decode: false
  - name: text
    dtype: string
  - name: source
    dtype: string
  - name: id
    dtype: string
  splits:
  - name: val
    num_bytes: 169694428.0
    num_examples: 1000
  download_size: 163153426
  dataset_size: 169694428.0
- config_name: kannada
  features:
  - name: audio
    dtype:
      audio:
        sampling_rate: 16000
        decode: false
  - name: text
    dtype: string
  - name: source
    dtype: string
  - name: id
    dtype: string
  splits:
  - name: val
    num_bytes: 180989518.0
    num_examples: 1000
  download_size: 176065039
  dataset_size: 180989518.0
- config_name: kashmiri
  features:
  - name: audio
    dtype:
      audio:
        sampling_rate: 16000
        decode: false
  - name: text
    dtype: string
  - name: source
    dtype: string
  - name: id
    dtype: string
  splits:
  - name: val
    num_bytes: 176844018.0
    num_examples: 1000
  download_size: 171833755
  dataset_size: 176844018.0
---

# am-pranav/stt-unified-bench

Private, language/locale-partitioned mini-benchmark for STT models.
Each **subset is a dataset config** (e.g., `en`, `de`, `en_indian_accent`, `hi_in`, `en_us`) with a single split **`val`**.
Audio is staged at 16 kHz and stored in-repo for reproducibility.

**Schema**
- `audio`  : `Audio(sampling_rate=16000, decode=False)`
- `text`   : reference transcription
- `lang`   : implied by dataset config name
- `source` : upstream dataset tag
- `id`     : source-stable id

> ⚠️ For internal evaluation only. Respect upstream licenses.