README.md

metadata

pretty_name: SaarAI ASR Benchmark Outputs
language:
  - amh
  - kin
  - nde
  - sna
  - zul
task_categories:
  - automatic-speech-recognition
tags:
  - asr
  - benchmark
  - model-outputs
  - saarai
configs:
  - config_name: amharic
    data_files:
      - split: clean_studio
        path: data/amharic/clean_studio__*.jsonl
  - config_name: kinyarwanda
    data_files:
      - split: test
        path: data/kinyarwanda/test__*.jsonl
  - config_name: ndebele
    data_files:
      - split: test
        path: data/ndebele/test__*.jsonl
  - config_name: shona
    data_files:
      - split: test
        path: data/shona/test__*.jsonl
  - config_name: zulu
    data_files:
      - split: test
        path: data/zulu/test__*.jsonl

SaarAI ASR Benchmark Outputs

Contents

language	splits	files	utterances
amharic	clean_studio	4	63860
kinyarwanda	test	9	90684
ndebele	test	9	17397
shona	test	13	22737
zulu	test	40	67914