Hugging Face's logo

STMicroelectronics
/
stft_tcnn

Audio-to-Audio
Model card Files Community
FBAGSTM commited on
Commit
f0b1fa1
·
verified ·
1 Parent(s): 883e0e5

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +4 -4
README.md CHANGED
@@ -62,7 +62,7 @@ Measures are done with default STEDGEAI configuration with enabled input / outpu
62
  ### Reference **NPU** memory footprint
63
  |Model | Dataset | Format | Resolution | Series | Internal RAM | External RAM | Weights Flash | STM32Cube.AI version | STEdgeAI Core version |
64
  |----------|------------------|--------|-------------|------------------|------------------|---------------------|-------|----------------------|-------------------------|
65
- | [STFT-TCNN Medium](https://github.com/STMicroelectronics/stm32ai-modelzoo/speech_enhancement/stft_tcnn/ST_pretrainedmodel_public_dataset/valentini/stft_tcnn_medium_sigmoid_257x40_qdq_int8.onnx) | valentini | Int8 | 257x40 | STM32N6 | 100.09 | 0.0 | 1626.86 | 10.0.0 | 2.0.0 |
66
 
67
  ### Reference **NPU** inference time
68
 
@@ -75,7 +75,7 @@ You can expect significant improvements once this issue is resolved.
75
 
76
  | Model | Dataset | Format | Resolution | Board | Execution Engine | Inference time (ms) | Inf / sec | STM32Cube.AI version | STEdgeAI Core version |
77
  |--------|------------------|--------|-------------|------------------|------------------|---------------------|-------|----------------------|-------------------------|
78
- | [STFT-TCNN medium](https://github.com/STMicroelectronics/stm32ai-modelzoo/speech_enhancement/stft_tcnn/ST_pretrainedmodel_public_dataset/valentini/stft_tcnn_medium_sigmoid_257x40_qdq_int8.onnx) | valentini | Int8 | 257x40 | STM32N6570-DK | NPU/MCU | 53.76 | 18.6 | 10.0.0 | 2.0.0 |
79
 
80
 
81
  ### Metrics on the Valentini dataset
@@ -90,8 +90,8 @@ We report five metrics :
90
 
91
  | Model | Format | Resolution | PESQ | STOI | SNR | SI-SNR | Waveform MSE |
92
  |-------|--------|------------|------|------|-----|--------|--------------|
93
- | [STFT-TCNN Medium](https://github.com/STMicroelectronics/stm32ai-modelzoo/speech_enhancement/stft_tcnn/ST_pretrainedmodel_public_dataset/valentini/stft_tcnn_medium_sigmoid_257xsl_float.onnx) | float32 | 257x? | 2.480 | 0.931 | 18.190 | 18.104 | 1.136e-4 |
94
- | [STFT-TCNN Medium](https://github.com/STMicroelectronics/stm32ai-modelzoo/speech_enhancement/stft_tcnn/ST_pretrainedmodel_public_dataset/valentini/stft_tcnn_medium_sigmoid_257xsl_qdq_int8.onnx) | int8 | 257x? | 2.372 | 0.932 | 18.190 | 18.100 | 1.109e-4 |
95
 
96
  ### Limitations
97
 
 
62
  ### Reference **NPU** memory footprint
63
  |Model | Dataset | Format | Resolution | Series | Internal RAM | External RAM | Weights Flash | STM32Cube.AI version | STEdgeAI Core version |
64
  |----------|------------------|--------|-------------|------------------|------------------|---------------------|-------|----------------------|-------------------------|
65
+ | [STFT-TCNN Medium](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/speech_enhancement/stft_tcnn/ST_pretrainedmodel_public_dataset/valentini/stft_tcnn_medium_sigmoid_257x40_qdq_int8.onnx) | valentini | Int8 | 257x40 | STM32N6 | 100.09 | 0.0 | 1626.86 | 10.0.0 | 2.0.0 |
66
 
67
  ### Reference **NPU** inference time
68
 
 
75
 
76
  | Model | Dataset | Format | Resolution | Board | Execution Engine | Inference time (ms) | Inf / sec | STM32Cube.AI version | STEdgeAI Core version |
77
  |--------|------------------|--------|-------------|------------------|------------------|---------------------|-------|----------------------|-------------------------|
78
+ | [STFT-TCNN medium](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/speech_enhancement/stft_tcnn/ST_pretrainedmodel_public_dataset/valentini/stft_tcnn_medium_sigmoid_257x40_qdq_int8.onnx) | valentini | Int8 | 257x40 | STM32N6570-DK | NPU/MCU | 53.76 | 18.6 | 10.0.0 | 2.0.0 |
79
 
80
 
81
  ### Metrics on the Valentini dataset
 
90
 
91
  | Model | Format | Resolution | PESQ | STOI | SNR | SI-SNR | Waveform MSE |
92
  |-------|--------|------------|------|------|-----|--------|--------------|
93
+ | [STFT-TCNN Medium](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/speech_enhancement/stft_tcnn/ST_pretrainedmodel_public_dataset/valentini/stft_tcnn_medium_sigmoid_257xsl_float.onnx) | float32 | 257x? | 2.480 | 0.931 | 18.190 | 18.104 | 1.136e-4 |
94
+ | [STFT-TCNN Medium](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/speech_enhancement/stft_tcnn/ST_pretrainedmodel_public_dataset/valentini/stft_tcnn_medium_sigmoid_257xsl_qdq_int8.onnx) | int8 | 257x? | 2.372 | 0.932 | 18.190 | 18.100 | 1.109e-4 |
95
 
96
  ### Limitations
97