Pl Asr Pelcra For Bigos: Polish Speech Subsets for ASR Benchmarking

Subsets of the SpokesMix, SpokesBiz, and Diabiz corpora processed into the BIGOS (Benchmark Intended Grouping of Open Speech) format. The data was contributed by the organization 'pelcra' and was last updated on October 26, 2024. The corpora provide spontaneous, conversational speech and phone-based customer interactions in Polish.

Use Cases

• Benchmarking Polish ASR model performance based on the described spontaneous and conversational speech.
• Training ASR systems for customer service applications based on the phone-based interaction data.
• Evaluating ASR robustness across different acoustic conditions and speaker demographics as suggested by the benchmark's intended scope.

Strengths

• Includes spontaneous and conversational speech, which is often more challenging for ASR than scripted audio.
• Contains phone-based customer interactions, providing a domain-specific use case.
• Aims to be a comprehensive public benchmark for Polish ASR, as stated in the description.

Limitations

• Description metadata is limited; actual data quality, size, and structure require manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file size are unknown, which may limit suitability assessment.

Provenance

Source

Subsets derived from the SpokesMix, SpokesBiz, and Diabiz corpora.

Collection Method

Processed into the BIGOS benchmark format.

Time Range

null

Freshness

Last updated 2024-10-26 15:15:33; freshness should be verified.

Geography

null