Librispeech: Difference between revisions

From MSc Voice Technology
Jump to navigation Jump to search
No edit summary
No edit summary
Line 19: Line 19:
* The dataset is split into 3 sections with 100.6, 363.6, 496.7 hours of speech.
* The dataset is split into 3 sections with 100.6, 363.6, 496.7 hours of speech.
* The gender ratio of speakers is about half and half.
* The gender ratio of speakers is about half and half.
* The size of each audio file is around
* The size of most audio files is less than 500kb.
|}
|}



Revision as of 17:08, 27 September 2023

Dataset name Type of data Annotation remarks License Metadata
LibriSpeech FLAC
  • The audio file and the transcription file are stored under the same directory.
  • There are multiple audio files and only one txt file which is the transcription.
  • And in the transcription file, the transcription for each audio is in separate lines.
Creative Commons Attribution 4.0 International License.
  • Language: English
  • The dataset is split into 3 sections with 100.6, 363.6, 496.7 hours of speech.
  • The gender ratio of speakers is about half and half.
  • The size of most audio files is less than 500kb.

Comparison Between Transcription and Original Transcripts

Material source: LibriSpeech-dev-clean-1272-128104

Google WER Whisper WER
1272-128104-0000.flac 23.52941176 17.64705882
1272-128104-0001.flac 40 20
1272-128104-0002.flac 21.875 12.5
1272-128104-0003.flac 29.16666667 16.66666667
1272-128104-0004.flac 33.82352941 26.47058824
1272-128104-0005.flac 16.66666667 16.66666667
1272-128104-0006.flac 50 33.33333333
1272-128104-0007.flac 5.263157895 21.05263158
1272-128104-0008.flac 45.45454545 45.45454545
1272-128104-0009.flac 37.20930233 13.95348837
1272-128104-0010.flac 26.66666667 20
1272-128104-0011.flac 52.94117647 14.70588235
1272-128104-0012.flac 33.33333333 33.33333333
1272-128104-0013.flac 33.33333333 16.66666667
1272-128104-0014.flac 75 25