🇱🇹 Lithuanian ASR Leaderboard (Local Results)
Overall (averages)
Model | Avg WER (all) | Avg CER (all) | Avg WER (excl in) | Avg CER (excl in) |
|---|---|---|---|---|
whisper-large-v3-turbo-lithuanian-lithuania-0.06-cer-filtered-ggml.bin | 1.1824999999999999 | 0.6599999999999999 | 0.38000000000000006 | 0.24333333333333332 |
CV13 (WER, CER)
Model | WER | CER |
|---|---|---|
whisper-large-v3-turbo-lithuanian-lithuania-0.06-cer-filtered-ggml.bin | 0.37 | 0.04 |
CV17 (WER, CER)
Model | WER | CER |
|---|---|---|
whisper-large-v3-turbo-lithuanian-lithuania-0.06-cer-filtered-ggml.bin | 0.37 | 0.04 |
LIEPA2 (WER, CER)
Model | WER | CER |
|---|---|---|
whisper-large-v3-turbo-lithuanian-lithuania-0.06-cer-filtered-ggml.bin | 0.66 | 0.12 |
LITHUANIA (WER, CER)
Model | WER | CER |
|---|---|---|
whisper-large-v3-turbo-lithuanian-lithuania-0.06-cer-filtered-ggml.bin | 0.49 | 0.13 |
Evaluation
- GPU: 1 × NVIDIA GeForce RTX 4090
📚 Models
- Whisper Large V2 Lithuanian (GGML)
- Whisper Large V3 Turbo Lithuanian 0.06 CER Filtered (GGML)
- Whisper Base Lithuanian (GGML)
🗂 Datasets
- Lithuania 0.06 CER Filtered
- Liepa-2
- Common Voice 13.0 (Lithuanian subset)
- Common Voice 17.0 (Lithuanian subset)
Tip: add new JSON files to asr_data/results/ and click Reload to refresh the tables.