Voxceleb1 Too Short Utts contains audio segments from the original VoxCeleb1 dataset. The dataset was created by s3prl and last updated on Hugging Face in July 2022. It focuses on utterances below a certain duration threshold.
Use Cases
- Train speaker embedding models using short-duration audio utterances.
- Benchmark speaker verification systems on challenging, brief speech segments.
- Analyze the impact of utterance length on speaker recognition accuracy.
- Develop models robust to variable-length audio inputs for real-world applications.
Strengths
- Derived from the well-known VoxCeleb1 dataset containing over 100,000 utterances.
- Focuses on a specific, challenging subset of short-duration speech data.
Limitations
- Specific row count and audio duration thresholds are unknown.
- Limited to celebrity speech data, which may not represent general population voices.
- Data is several years old, with the last update in 2022.
Provenance
- Source
- VoxCeleb1 dataset.
- Collection Method
- Subset extraction of short utterances from the original VoxCeleb1 audio files.
- Time Range
- null
- Freshness
- Last updated in 2022.
- Geography
- null