Keyword Spotting on LibriPhrase Easy (LPE)
0.45EERDMA-KWS(2)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DMA-KWS(2)# Params=4.1M, PT (hrs)=1460†, FT (hrs)=14602026.05 | 0.45 | 99.98 | |
| DMA-KWS(1)# Params=4.1M, PT (hrs)=1460†, FT (hrs)=14602026.05 | 0.52 | 99.98 | |
| PLCL# Params=3.9M2026.03 | 0.57 | 99.97 | |
| PLCL@ATVenue=ICASSP 25, # Params=40M, PT (hrs)=680,000†, FT (hrs)=4602026.05 | 0.57 | 99.97 | |
| PLCLAugmentation strategy=method-specific2026.06 | 0.57 | 99.97 | |
| DMA-KWS(2)# Params=4.1M, PT (hrs)=460†, FT (hrs)=4602026.05 | 0.59 | 99.96 | |
| ProKWS# Params=2.9M2026.03 | 0.63 | 99.96 | |
| MM-KWS# Params=3.9M2026.03 | 0.68 | 99.95 | |
| MM-KWS@ATVenue=IS 24, # Params=3.9M, PT (hrs)=0, FT (hrs)=460+417‡2026.05 | 0.68 | 99.95 | |
| MM-KWSAugmentation strategy=method-specific2026.06 | 0.68 | 99.95 | |
| DMA-KWS(1)# Params=4.1M, PT (hrs)=460†, FT (hrs)=4602026.05 | 0.72 | 99.96 | |
| MM-KWS@TVenue=IS 24, # Params=3.9M, PT (hrs)=0, FT (hrs)=460+417‡2026.05 | 0.82 | 99.94 | |
| W-CTCVenue=IS 25, # Params=3.6M, PT (hrs)=460†, FT (hrs)=4602026.05 | 0.91 | 99.95 | |
| AdaKWS-SmallVenue=ICASSP 24, # Params=109M, PT (hrs)=1800†, FT (hrs)=4602026.05 | 1.21 | 99.82 | |
| PLCL@TVenue=ICASSP 25, # Params=40M, PT (hrs)=680,000†, FT (hrs)=4602026.05 | 1.21 | 99.95 | |
| AdaKWS-SmallAugmentation strategy=method-specific2026.06 | 1.21 | 99.82 | |
| ADMLVenue=IS 25, # Params=1.8M, PT (hrs)=4600*, FT (hrs)=02026.05 | 1.33 | 99.86 | |
| MATETraining Dataset=King-ASR-066, K=32026.01 | 1.38 | 99.86 | |
| ADMLTraining Dataset=King-ASR-066, Components=w/o SF22026.01 | 1.43 | 99.84 | |
| RPLTraining Dataset=King-ASR-0662026.01 | 1.54 | 99.84 | |
| AdaKWS-Tiny# Params=15M2026.03 | 1.61 | 99.8 | |
| AdaKWS-TinyVenue=ICASSP 24, # Params=15M, PT (hrs)=1800†, FT (hrs)=4602026.05 | 1.61 | 99.8 | |
| CED# Params=3.6M2026.03 | 1.7 | 99.84 | |
| CEDVenue=ICASSP 24, # Params=3.6M, PT (hrs)=460†, FT (hrs)=4602026.05 | 1.7 | 99.84 | |
| CEDAugmentation strategy=method-specific2026.06 | 1.7 | 99.84 | |
| SLiCKVenue=ICASSP 25, # Params=0.6M, PT (hrs)=460†, FT (hrs)=4602026.05 | 1.78 | 99.82 | |
| KFC-KWSAugmentation strategy=modality dropout2026.06 | 1.94 | 99.81 | |
| PhonMatchNetTraining Dataset=King-ASR-0662026.01 | 2.33 | 99.7 | |
| PhonMatchNet# Params=0.7M2026.03 | 2.8 | 99.29 | |
| PhonMatchNetVenue=IS 23, # Params=0.7M, PT (hrs)=110,000*, FT (hrs)=4602026.05 | 2.8 | 99.29 | |
| CMCDTraining Dataset=King-ASR-0662026.01 | 4.2 | 99.12 | |
| Whisper-Large# Params=1550M2026.03 | 5.33 | 97.54 | |
| SynaspotEnrollment modality (EM)=Text + Audio, Params=0.9M2025.12 | 5.77 | 97.34 | |
| CTCATVenue=IS 24, # Params=0.2M, PT (hrs)=0, FT (hrs)=4602026.05 | 6.06 | 98.32 | |
| Synaspot-ATEnrollment modality (EM)=Text, Params=0.9M2025.12 | 7.07 | 97.17 | |
| EMKWS# Params=3.7M2026.03 | 7.36 | 97.83 | |
| EMKWSVenue=IS 23, # Params=3.7M, PT (hrs)=460†, FT (hrs)=4602026.05 | 7.36 | 97.83 | |
| Whisper-Small# Params=224M2026.03 | 8.14 | 95.92 | |
| CMCDEnrollment modality (EM)=Text, Params=0.7M2025.12 | 8.42 | 96.7 | |
| CMCD# Params=0.7M2026.03 | 8.42 | 96.7 | |
| CMCDVenue=IS 22, # Params=0.7M, PT (hrs)=0, FT (hrs)=4602026.05 | 8.42 | 96.7 | |
| CLADEnrollment modality (EM)=Text, Params=2.2M2025.12 | 8.65 | 97.03 | |
| CLAD# Params=2.2M2026.03 | 8.65 | 97.03 | |
| CLADVenue=ICASSP 24, # Params=3.6M, PT (hrs)=460*, FT (hrs)=02026.05 | 8.65 | 97.03 | |
| Synaspot-AAEnrollment modality (EM)=Audio, Params=0.9M2025.12 | 8.85 | 96.15 | |
| InfoNCEEnrollment modality (EM)=Text, Params=2.2M2025.12 | 8.99 | 96.85 | |
| Whisper-Tiny# Params=39M2026.03 | 17.31 | 89.19 | |
| SoftTripeEnrollment modality (EM)=Text, Params=N/A2025.12 | 28.74 | 78.74 | |
| DONUTVenue=ICASSP 21, # Params=-, PT (hrs)=0, FT (hrs)=4602026.05 | 28.74 | 78.74 | |
| TripletEnrollment modality (EM)=Text, Params=0.6M2025.12 | 32.75 | 63.53 | |
| TripletVenue=IS 19, # Params=-, PT (hrs)=0, FT (hrs)=4602026.05 | 32.75 | 63.53 |