Dialogue System Challenge Automatic Speech Recognition Data

Dstc23 Asr is a dataset for the Dialogue System Technology Challenge (DSTC) focused on automatic speech recognition. The dataset was created by congpt and uploaded to Hugging Face in April 2021. It is designed for evaluating and training models on transcribing spoken dialogue.

Use Cases

• Train ASR models to transcribe audio from spoken dialogue interactions.
• Benchmark ASR system performance on conversational speech data from the DSTC challenge.
• Fine-tune speech-to-text models using domain-specific dialogue audio and corresponding transcriptions.

Strengths

• Associated with the established DSTC research challenge series.
• Focuses on the specific domain of conversational speech recognition.

Limitations

• Unknown dataset size, row count, and specific audio characteristics.
• Data is from 2021 and may not reflect recent speech patterns or acoustic environments.
• Geographic and demographic coverage of speakers is unspecified.

Provenance

Source

Dialogue System Technology Challenge (DSTC)

Collection Method

Likely collected or curated for the DSTC23 challenge track.

Freshness

Last updated in April 2021.