MultiSpeech
Collection
MULTISPEECHQA, a multilingual speech-language instruction fine-tuning
dataset and MULTISPEECH-BENCH, a multilingual, multitask speech processing bench
•
1 item
•
Updated
Datasets:
Dataset Viewer
Search is not available for this dataset
audio
audioduration (s) 1
1
|
|---|
End of preview. Expand
in Data Studio
test_multispeech
Multilingual multitask audio dataset (dummy data).
Tasks: asr, sqa, ast
Languages: en, fr, es, ar
Splits: train, validation, test
Usage:
from datasets import load_dataset
# Public-style (Hub)
ds = load_dataset("yourname/test_multispeech", "asr_fr", split="train")
# Local-style
ds = load_dataset("./test_multispeech", task="sqa", language="en", split="train")
- Downloads last month
- 196
Size of downloaded dataset files:
11.5 MB
Size of the auto-converted Parquet files:
16.8 kB
Number of rows:
360