Intent Classification on SLURP
Accuracy (%)
- 2020-11-26 — Multi-SLURP: Accuracy (%) 78.33
- 2021-11-04 — Partially Fine-tuned HuBERT: Accuracy (%) 87.51
- 2023-04-13 — TDT 0-8: Accuracy (%) 90.07
| Rank | Model | Accuracy (%) | Extra Training Data | Paper | Code | Year |
|---|---|---|---|---|---|---|
| 1 | TDT 0-8 | 90.07 | Efficient Sequence Transduction by Jointly Predicting Tokens and Durations | NVIDIA/NeMo · chimechallenge/C8DASR-Baseline-NeMo · kehanlu/Nemo · +1 | 2023 | |
| 2 | Partially Fine-tuned HuBERT | 87.51 | A Fine-tuned Wav2vec 2.0/HuBERT Benchmark For Speech Emotion Recognition, Speaker Verification and Spoken Language Understanding | 2021 | ||
| 3 | Multi-SLURP | 78.33 | ✓ | SLURP: A Spoken Language Understanding Resource Package | pswietojanski/slurp | 2020 |
| 4 | Finstreder (Conformer) | 53.11 | ✓ | Finstreder: Simple and fast Spoken Language Understanding with Finite State Transducers using modern Speech-to-Text models | Jaco-Assistant/Jaco-Master · Jaco-Assistant/finstreder | 2022 |
| 5 | Finstreder (Quartznet) | 43.15 | ✓ | Finstreder: Simple and fast Spoken Language Understanding with Finite State Transducers using modern Speech-to-Text models | Jaco-Assistant/Jaco-Master · Jaco-Assistant/finstreder | 2022 |