Datasets:
Tasks:
Automatic Speech Recognition
Sub-tasks:
keyword-spotting
Size:
10K<n<100K
ArXiv:
Tags:
speech-recognition
License:
To All who load failed
#32
by
davidsh
- opened
if you use the default code:
from datasets import load_dataset, Audio
dataset = load_dataset("PolyAI/minds14", "en-US", split="train")
you can get error (my datasets version is 3.3.2):
File "~/.venv/lib/python3.12/site-packages/datasets/load.py", line 1886, in load_dataset_builder
builder_instance: DatasetBuilder = builder_cls(
^^^^^^^^^^^^
TypeError: 'NoneType' object is not callable
change code to get the right One:
dataset = load_dataset("PolyAI/minds14", data_dir="en-US", split="train", revision="refs/convert/parquet")
Importtant:
- revision: the branch that has parquet
- data_dir: the data path in your specified revision
- split: virtual split or real dir
Best Way:
downgrade your datasets version to 2.20.0
pip3 install datasets==2.20.0
then the following will be OK!
from datasets import load_dataset, Audio
dataset = load_dataset("PolyAI/minds14", "en-US", split="train")