Datasets:

rustam1221
/

uzbek-asr-benchmark-spontaneous

Tasks:

Automatic Speech Recognition

Languages:

Size:

Tags:

spontaneous-speech

License:

Dataset card Data Studio Files Files and versions

uzbek-asr-benchmark-spontaneous

2.48 MB

Ctrl+K

Ctrl+K

1 contributor

History: 149 commits

rustam1221's picture

uzasr-base-v0.3.0 on v3: 9.43% WER / 2.96% CER (prod path), predictions

364f5a5 verified 8 days ago

predictions
uzasr-base-v0.3.0 on v3: 9.43% WER / 2.96% CER (prod path), predictions 8 days ago
.gitattributes

2.5 kB
initial commit 3 months ago
README.md

11.4 kB
uzasr-base-v0.3.0 on v3: 9.43% WER / 2.96% CER (prod path), predictions 8 days ago
adjudication_v3.jsonl

53 kB
Benchmark v3: listening adjudication corrected 144 clips; predictions and summary re-scored; wer_gate in scoring.py 9 days ago
changes_v3.jsonl

99.8 kB
Benchmark v3: listening adjudication corrected 144 clips; predictions and summary re-scored; wer_gate in scoring.py 9 days ago
corrections.jsonl

40.7 kB
etalon tahriri 3 months ago
etalon.jsonl

134 kB
Upload etalon.jsonl with huggingface_hub 3 months ago
exclude_rows.json

4.66 kB
train exclusion map: 315 etalon source_rows -> parquet file+row 3 months ago
final_etalon.jsonl

138 kB
Benchmark v3: listening adjudication corrected 144 clips; predictions and summary re-scored; wer_gate in scoring.py 9 days ago
final_etalon_v2.jsonl

138 kB
Benchmark v3: listening adjudication corrected 144 clips; predictions and summary re-scored; wer_gate in scoring.py 9 days ago
reconstruct_benchmark_audio.py

2.99 kB
docs: English cards, 600M results, standalone scripts 2 months ago
scoring.py

3.6 kB
Benchmark v3: listening adjudication corrected 144 clips; predictions and summary re-scored; wer_gate in scoring.py 9 days ago
source_map.json

18 kB
docs: English cards, 600M results, standalone scripts 2 months ago