Keyword Spotting On Google Speech Commands

Metrics

Google Speech Commands V2 35

Results

Performance results of various models on this benchmark

Model Name	Google Speech Commands V2 35	Paper Title	Repository
HTS-AT	98.0	HTS-AT: A Hierarchical Token-Semantic Audio Transformer for Sound Classification and Detection	-
BC-ResNet-8	-	Broadcasted Residual Learning for Efficient Keyword Spotting	-
WaveFormer	99.1	Work in Progress: Linear Transformers for TinyML	-
ImportantAug	95	ImportantAug: a data augmentation agent for speech	-
TripletLoss-res15	97.0	Learning Efficient Representations for Keyword Spotting with Triplet Loss	-
LSTM	-	Hello Edge: Keyword Spotting on Microcontrollers	-
DenseNet-BiLTSM	-	Effective Combination of DenseNet andBiLSTM for Keyword Spotting	-
GRU	-	Hello Edge: Keyword Spotting on Microcontrollers	-
Attention RNN	93.9	A neural attention model for speech command recognition	-
MatchboxNet-3x2x64	-	MatchboxNet: 1D Time-Channel Separable Convolutional Neural Network Architecture for Speech Commands Recognition	-
TDNN	-	Efficient keyword spotting using time delay neural networks	-
End-to-end KWS model	-	End-to-end Keyword Spotting using Neural Architecture Search and Quantization	-
LSTM	-	Multi-layer Attention Mechanism for Speech Keyword Recognition	-
DNN	-	Hello Edge: Keyword Spotting on Microcontrollers	-
Basic LSTM	-	Hello Edge: Keyword Spotting on Microcontrollers	-
Audio Spectrogram Transformer	98.11	AST: Audio Spectrogram Transformer	-
QNN	98.60	Towards on-Device Keyword Spotting using Low-Footprint Quaternion Neural Models
TC-ResNet14-1.5	-	Temporal Convolution for Real-time Keyword Spotting on Mobile Devices	-
SSAMBA	97.4	SSAMBA: Self-Supervised Audio Representation Learning with Mamba State Space Model	-
KWT-1	96.95±0.14	Keyword Transformer: A Self-Attention Model for Keyword Spotting	-

0 of 42 row(s) selected.