Datasets:

Modalities:
Audio
Text
Formats:
parquet
ArXiv:
License:

Corrupt audio sample in "Telugu - train" split

#5
by skesiraju - opened
from datasets import load_dataset
indic_voices = load_dataset("ai4bharat/IndicVoices", "telugu")
indic_voices['train']['audio_filepath'][64251].get_all_samples()

Throws RuntimeError: Could not receive frame from decoder: Invalid data found when processing input
This is just for the sample at index 64251.

The metadata is accessible

indic_voices['train']['audio_filepath'][64251].metadata

AudioStreamMetadata:
  duration_seconds_from_header: 1.792
  begin_stream_seconds_from_header: 0.0
  bit_rate: 0.0
  codec: flac
  stream_index: 0
  sample_rate: 16000
  num_channels: 1
  sample_format: s16
AI4Bharat org

Updated, please check

tahirjm changed discussion status to closed

Sign up or log in to comment