Polish Parliamentary Speeches Audio Collection

Comprising 97 hours of parliamentary speeches from Poland. The audio is stored in .wav format and was published on the ClarinPL website.

Use Cases

• Train an automatic speech recognition model on 97 hours of Polish parliamentary speech audio.
• Analyze speech patterns and durations across different parliamentary sessions using the audio files.
• Convert .wav files to float32 arrays for acoustic feature extraction and analysis.

Strengths

• 97 hours of audio data provides substantial material for speech analysis.
• Audio is stored in the .wav format, a standard lossless format for audio processing.
• Annotations are expert-generated, indicating a level of quality control.

Limitations

• The dataset size is categorized as 1K<n<10K, which is relatively small for some machine learning tasks.
• The audio is not pre-converted to float32 arrays, requiring an extra processing step for typical ML workflows.
• Limited metadata is available, as specific columns and sample data are unknown.

Provenance

Source

ClarinPL website.

Collection Method

Collection of published parliamentary speeches.

Freshness

Last updated on 2023-01-22.

Geography

Poland (inferred from language 'pl').