from huggingface_hub import login import pandas as pd from datasets import Dataset, Audio, Sequence, Value import ast # login() # This will ask for your token, which you can get from https://huggingface.co/settings/tokens repo_id = "pratikstar/doesitchug" # Replace 'your-username' with your Hugging Face username # Load the CSV file file_path = "./updated_timbre_dataset.csv" df = pd.read_csv(file_path) df['Label-Annotations'] = df['Label-Annotations'].apply(lambda x: ast.literal_eval(x) if isinstance(x, str) else x) print(len(df)) print(df["Clip-A"][0]) # print(df["Clip A"]) # Convert the pandas DataFrame to a Hugging Face Dataset hf_dataset = Dataset.from_pandas(df) # Specify that 'clip_a' and 'clip_b' columns are audio columns hf_dataset = hf_dataset.cast_column("Clip-A", Audio(sampling_rate=44100)) hf_dataset = hf_dataset.cast_column("Clip-B", Audio(sampling_rate=44100)) hf_dataset = hf_dataset.cast_column("Label-Annotations", Sequence(Value(dtype='string'))) print(hf_dataset[0]) # Push the dataset to the Hub hf_dataset.push_to_hub(repo_id)